{ "days": [ { "date": "2026-06-16", "plenary_talks": [ { "talk_id": "invited:Ulrich S. Schubert", "title": "Combining online characterization and synthetic robots – On the road to self-driving labs", "speaker": "Ulrich S. Schubert", "talk_type": "plenary", "length": 1, "start_time": "09:15", "end_time": "09:45", "abstract": "Robotic synthesis platforms revolutionized materials research starting in the late 1990s by enabling the parallel and automated execution of numerous experiments. In polymer chemistry, these platforms were employed to systematically explore polymerization parameters, such as temperature, monomer-to-initiator ratios, and reaction kinetics, to yield comprehensive material libraries for structure-property investigations. Yet, rapid characterization of these libraries proved challenging, as reliance on offline analytical methods created a persistent bottleneck that slowed down discovery workflows.\n\nWhile substantial efforts focused on developing new polymerization techniques (e.g., living or controlled polymerization methods) and refining existing ones, addressing the analytical bottleneck received comparatively less attention. Meanwhile, the impact of offline sampling on reaction processes was studied, leading to specialized methods that minimized sampling-induced disturbances. For a time, these compromises persisted. With the advent of dedicated tools, such as compact benchtop NMR spectrometers, the analytical bottleneck was overcome. These new devices enabled workflows for real-time studies on purification kinetics, e-fuel formation, energy storage or catalytic water splitting.\n\nIn recent years, diverse sensors have complemented benchtop NMR, vastly expanding online characterization capabilities. In situ spectropho-tometric sensors, for instance, now provide continuous monitoring capabilities for catalytic hydrogen evolution reactions, delivering real-time feedback, which could guide iterative optimization. These developments pave the way for self-driving labs across scientific domains, including catalyst optimization and battery research, where robotics, analytics, and machine learning converge to accelerate materials discovery.\n\nThe current contribution will highlight the development from using high-throughput experimentation towards the utilizat", "speaker_profile_url": "https://www.schubert-group.uni-jena.de/en/1480/about-prof-dr-ulrich-s-schubert", "section": [ "AI for Materials Science", "AI for Chemistry", "Self-Driving Labs" ] }, { "talk_id": "invited:Tommaso Dorigo", "title": "The second AI revolution in fundamental science", "speaker": "Tommaso Dorigo", "talk_type": "plenary", "length": 1, "start_time": "09:45", "end_time": "10:15", "abstract": "The success of automatic methods for image classification in 2012 mark a phase transition in the performance of machine learning algorithms; those developments led to a revolution in the way the extraction of information from complex data is operated in fundamental science experiments. A second AI-powered revolution is under way now, thanks to the development of more advanced, powerful algorithms and methods; its target is the assistance of humans in the optimal design of scientific experiments. The large dimensionality of the parameter space of the design of large experiments, the stochasticity of the physical processes generating relevant data, and the complexity of the objective function of multi-target experiments can now be handled by hybrid optimization techniques, multi-modal systems, and advanced generative AI. In this presentation the state of the art of research in this thriving subject will be discussed.", "speaker_profile_url": "https://userswww.pd.infn.it/~dorigo", "section": [ "AI for Physics" ] }, { "talk_id": "invited:Curtis Berlinguette", "title": "Ada-Carbon: A self-driving laboratory to enable the lowest-cost pathway to scalable $\\ce{CO_2}$-to-fuels conversion", "speaker": "Curtis Berlinguette", "talk_type": "plenary", "length": 1, "start_time": "11:00", "end_time": "11:30", "abstract": "Carbon capture and utilization pathways require that $\\ce{CO_2}$ captured from the atmosphere (or a point source) be released from the sorbent, and that the sorbent be recycled to capture additional $\\ce{CO_2}$. Alkaline solutions such as KOH are effective at capturing $\\ce{CO_2}$ through reactions that form (bi)carbonates, but the recovery of $\\ce{CO_2}$ gas and hydroxide before $\\ce{CO_2}$ electrolysis requires energy-intensive steps. We solved this problem by designing an electrochemical reactor that converts bicarbonate “reactive carbon capture solutions” into carbon-containing products. In this presentation, I will show how this reactor couples $\\ce{CO_2}$ conversion with upstream carbon capture. Not only does this reactor bypass the expensive step of liberating $\\ce{CO_2}$ from the sorbent, but it can also perform better than the reactors fed with gaseous $\\ce{CO_2}$.\n\nThe emergent challenge is how to balance $\\ce{CO_2}$ capture with $\\ce{CO_2}$ conversion. This is a multi-variant problem that we built an autonomous laboratory to solve. This self-driving lab is named Ada-Carbon. \n\nI will demonstrate how we reconfigured Ada, a self-driving laboratory designed for solar cells, into Ada-Carbon: a platform for discovering and optimizing electrochemical reactors that upgrade waste $\\ce{CO_2}$ into valuable fuels. We did this by using “flexible automation”. Flexible automation is the concept of creating reconfigurable automated experiments, enabling the integration of various experimental procedures into fully automated workflows. Importantly, flexible automation platforms can evolve with the changing needs of the experimentalist. This talk will show flexible automation in action.", "speaker_profile_url": "https://berlinguettegroup.com", "section": [ "AI for Materials Science", "AI for Chemistry", "Self-Driving Labs" ] }, { "talk_id": "invited:Laura Matz", "title": "Unlocking Precision Medicine | The Digital Ecosystem Powering Tomorrow's Therapies", "speaker": "Laura Matz", "talk_type": "plenary", "length": 1, "start_time": "12:15", "end_time": "12:45", "abstract": "Artificial intelligence is redefining drug discovery by weaving automation, AI/ML, and real-time data loops into the research pipeline, enabling AIDD to shorten the DMTA cycle by over 50% and accelerate precision therapies. The new standard in biology emphasizes human biology over animal models, with Microphysiological Systems using patient-specific cells to model disease, improving translatability and reducing animal testing. Data serves as the backbone of precision medicine, but only when it is trustworthy and shareable; Syntropy provides secure, scalable collaboration to unlock diverse biomedical data, expediting therapeutic discovery and improving patient outcomes. Merck KGaA’s cross-sector strengths - spanning Life Science, Healthcare, and Electronics - offer a unique competitive differentiator, enabling a holistic, end-to-end precision medicine ecosystem from molecule to patient. Together, these elements position a unified platform for faster, more reliable development of precision therapies and enhanced patient impact.", "speaker_profile_url": "https://www.merckgroup.com/en", "section": [ "AI for Medicine and Healthcare" ] } ], "keynote_talks": [ { "talk_id": "extra:keynote:opening:1", "title": "Opening", "speaker": "Kostya Novoselov & Alán Aspuru-Guzik", "talk_type": "extra_keynote", "length": 1, "start_time": "09:00", "end_time": "09:15", "speaker_profile_url": "https://en.wikipedia.org/wiki/Konstantin_Novoselov", "fixed_slot_start": "09:00", "fixed_slot_end": "09:15", "abstract": "" }, { "talk_id": "invited:Tejs Vegge", "title": "MaterialsCommons for Europe – SDLs and FAIR workflows for federated discovery of advanced materials", "speaker": "Tejs Vegge", "talk_type": "featured", "length": 1, "start_time": "10:15", "end_time": "10:30", "abstract": "Accelerating the discovery, synthesis and deployment of advanced materials increasingly relies on distributed self-driving laboratories (SDLs) [1] and materials acceleration platforms (MAPs) [2] that integrate simulation, ML, data and autonomous experimentation. The challenge is not only better ML models, but interoperable infrastructure that lets models, simulations, experiments and repositories operate as one reproducible discovery system. Here, we present MaterialsCommons, a large-scale federated, FAIR-by-design digital infrastructure for machine-actionable workflows for advanced materials across Europe [3].\nMaterialsCommons will provide a multi-node ecosystem with a unified single-entry point, semantic schemas and ontology-driven metadata that make materials and process data Findable, Accessible, Interoperable and Reusable (FAIR) from atomistic simulation to industrial validation (Figure 1). These capabilities are illustrated through federated materials-and-device co-optimization of batteries [4], interoperable DFT workflows that express open-circuit-voltage calculations through a common JSON/OPTIMADE input-output standard across AiiDA, PerQueue, Pipeline Pilot and SimStack and across CASTEP, GPAW, Quantum ESPRESSO and VASP [5], and FastCat, an AI-orchestrated MAP for multi-metal oxygen-evolution catalysts [6].\nCentral to MaterialsCommons is the Workflow-of-Workflows (WoW) concept, relying on dynamic workflow managers such as PerQueue [7,8] to call computational and experimental nodes across an interoperable network of distributed laboratories. The same logic extends from execution to representation learning, and here we introduce the concept of tomographic materials representation, where we interpret structures, properties and descriptors as complementary projections of an underlying material object [9], while deep tomography connects molecular representation learning to quantum-state tomography and shared latent representations from informationally rich observables [10].\nBy embedding these capabilities in a governance-backed, sustainably operated infrastructure, MaterialsCommons closes the loop between target structure, autonomous experimentation, reproducible protocol and information-aware AI. We will outline architectural principles, use cases spanning catalysis, batteries and nanomaterials, and a roadmap toward an AI-ready ecosystem for accelerated materials innovation.", "speaker_profile_url": "https://www.linkedin.com/in/tejs-vegge-b1aa291/", "section": [ "AI for Materials Science", "Self-Driving Labs" ] }, { "talk_id": "invited:Maria K. Y. Chan", "title": "Seeing the invisible in materials with AI", "speaker": "Maria K. Y. Chan", "talk_type": "featured", "length": 1, "start_time": "11:30", "end_time": "11:45", "abstract": "Advances in AI are rapidly transforming how we understand, characterize, and design materials and chemistry for energy applications including energy storage and catalysis. In this talk, I will discuss key challenges and some efforts in applying AI towards materials and chemistry, including integration of theory-guided modeling with AI/ML approaches to interpret complex experimental characterization data (from electron and x-ray microscopy to spectroscopy) enabling “seeing the invisible” at atomic scales. I discuss strategies for property prediction, autonomous experimentation, the extraction of microscopy and spectroscopy data from scientific literature, and the development of data standards and infrastructure that make experimental data AI-ready. The talk draws on research funded by the US Department of Energy such as a DOE Early Career award, Energy Storage Research Alliance (ESRA), Midwest Integrated Center for Computational Materials (MICCoM), and the Integrated Scientific Agentic AI for Catalysis (ISAAC) project under the Genesis Mission.", "speaker_profile_url": "https://www.anl.gov/profile/maria-k-chan", "section": [ "AI for Materials Science", "AI for Science", "Self-Driving Labs" ] }, { "talk_id": "extra:keynote:advances-in-automating-synthetic-chemistry-uot-and-tecan:1", "title": "Accelerating Hit Optimization with Automated Parallel Synthesis", "speaker": "Lesley Schultz, Santha Santhakumar (TECAN & Acceleration Consortium)", "talk_type": "sponsor", "length": 1, "start_time": "11:45", "end_time": "12:00", "abstract": "" }, { "talk_id": "extra:keynote:the-autonomous-materials-foundry-the-physical-infrastructure-for-ai-driven-materials-discovery-with-direct-atomic-layer-processing-dalp-technology:1", "title": "The Autonomous Materials Foundry: The Physical Infrastructure for AI-Driven Materials Discovery with Direct Atomic Layer Processing (DALP®) Technology", "speaker": "Maksym Plakhotnyuk, ATLANT 3D", "talk_type": "sponsor", "length": 1, "start_time": "12:45", "end_time": "13:00", "abstract": "The next frontier of artificial intelligence is not purely digital; it is physical. As AI models accelerate the discovery of new materials, a critical bottleneck remains: the ability to rapidly, precisely, and autonomously fabricate and validate those materials in the real world. This talk introduces the Autonomous Materials Foundry— physical infrastructure designed to close the loop between AI-driven discovery and atomic-scale manufacturing to test, validate and scale new materials.\nAt the core of this is Direct Atomic Layer Processing (DALP®), a breakthrough technology enabling digitally-defined, maskless, and direct fabrication of materials at the atomic scale. By integrating DALP® technology with AI models, real-time data feedback, and automated experimentation, the Autonomous Materials Foundry transforms materials development into a fully digitized, software-defined process.\nWe will explore how this approach enables rapid iteration cycles, decentralized production, and unprecedented control over material properties—unlocking new possibilities in semiconductors, quantum, energy systems, and beyond. The talk will also examine how such infrastructure can serve as a foundational layer for AI-native industries, enabling a shift from trial-and-error experimentation to predictive, self-optimizing materials engineering.\nUltimately, the Autonomous Materials Foundry represents a step toward a future where materials innovation is not only accelerated by AI, but autonomously executed in the physical world.", "speaker_profile_url": "https://atlant3d.com/" }, { "talk_id": "extra:keynote:launch-of-ai-for-science-accelerating-discovery-through-ai:1", "title": "Launch of AI for Science: Accelerating Discovery Through AI", "speaker": "Tan Chorh Chuan", "talk_type": "extra_keynote", "length": 1, "start_time": "13:00", "end_time": "13:15", "abstract": "Prof. Tan Chorh Chuan will be officially launching the AI-for-Science (AI4S) programme. The AI4S initiative is a Singapore national funding initiative by the National Research Foundation (NRF) that aims to advance the development of artificial intelligence tools to transform the pace and scope of scientific discovery. The programme brings together collaborative teams of AI and domain researchers, from leading institutions across Singapore and internationally, nurturing a next generation of bilingual scientists fluent in both AI and their respective scientific fields. Through its Challenge and Catalytic grant schemes, AI4S supports both large-scale, transformative research projects and creative, exploratory ideas at the frontier of science, including life sciences, materials science, agriculture, and quantum science.", "speaker_profile_url": "https://en.wikipedia.org/wiki/Tan_Chorh_Chuan", "fixed_slot_start": "13:00", "fixed_slot_end": "13:15" } ], "early_morning_chair_name": "Shi Xuan Leong", "early_morning_chair_profile_url": "https://www.leongshixuan.com/", "late_morning_chair_name": "Jacqueline Cole", "late_morning_chair_profile_url": "https://en.wikipedia.org/wiki/Jacqui_Cole", "midday_chair_name": "Kostya Novoselov", "time_blocks": [ { "block_name": "afternoon_session", "start_time": "14:45", "end_time": "16:15", "sessions": [ { "session_title": "AI for Chemistry and Materials Discovery", "rationale": "Physical Unified Device Architecture for AI-Assisted $\\ce{CO_2}$ Electrocatalysis, Accelerating Ammonia Decomposition Catalyst Discovery with AI, Towards ab-initio quality description of porous materials: Developing general Machine-Learned Potentials to simulate physical and adsorption properties of Metal-Organic Frameworks, Diffusion-Driven Generation of Novel Crystalline Materials with Target Optical Properties, Atom-in-molecule based quantum machine learning of defect formation energies, AI-guided experimental design of zirconium MOPs with The World Avatar for sustainable photocatalysis", "section": "AI for Chemistry", "breakout_venue_name": "Moor", "chair_name": "Berend Smit", "chair_profile_url": "https://epfl.ch/labs/lsmo/smit", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Yanwei Lum", "title": "Physical Unified Device Architecture for AI-Assisted $\\ce{CO_2}$ Electrocatalysis", "speaker": "Yanwei Lum", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "High-throughput $\\ce{CO_2}$ electrocatalysis requires reproducible experimental workflows that connect hardware, data, and decision-making. However, current laboratory systems are fragmented across heterogeneous instruments, vendor-specific data formats, and manual analysis procedures. Here, we present a Physical Unified Device Architecture (PUDA)-enabled infrastructure for AI-assisted $\\ce{CO_2}$ electrocatalysis research. PUDA functions as a loop-closure architecture that converts laboratory instruments and robotic platforms into command-line-accessible, Large Language Model (LLM)-invokable physical endpoints. Through a unified middleware layer, electrochemical workstations, chromatographic instruments, mass-flow controllers, robotic systems, and characterization tools can be connected to AI agents for controlled execution, data acquisition, and workflow orchestration. This is integrated with a centralized data-processing platform that automatically synchronizes raw experimental files, parses them into structured formats, and enables traceable analysis through natural-language instructions. The current implementation supports a full spectrum data processing techniques ranging from electrochemical workstation, gas chromatography, high-performance liquid chromatography, mass-flow controllers, nuclear magnetic resonance, X-ray photoelectron spectroscopy, ultraviolet-visible spectroscopy, X-ray diffraction, and X-ray absorption spectroscopy. In $\\ce{CO_2}$ electrocatalysis workflows, the system combines electrochemical, gas-product, liquid-product, and flow-rate data to perform calibration, concentration calculation, product quantification, and Faradaic efficiency analysis directly from raw files. This PUDA-enabled framework provides an AI-native experimental backbone for closed-loop catalyst screening and $\\ce{CO_2}$ electrocatalysis test bedding. By standardizing physical tool access and raw-data processing, it improves throughput, reproducibility, and data traceability, while laying the foundation for autonomous optimization of catalysts and electrolyzer systems.", "speaker_profile_url": "https://cde.nus.edu.sg/chbe/staff/lum-yanwei/", "section": [ "AI for Chemistry" ] }, { "talk_id": "oral:accelerating-ammonia-decomposition-catalyst-discovery-with-ai", "title": "Accelerating Ammonia Decomposition Catalyst Discovery with AI", "speaker": "Mathilde Franckel", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=za4lpuV7Cx", "openreview_id": "za4lpuV7Cx", "speaker_profile_url": "https://openreview.net/profile?id=~Mathilde_L._D._Franckel1" }, { "talk_id": "oral:towards-ab-initio-quality-description-of-porous-materials-developing-general-machine-learned-potentials-to-simulate-physical-and-adsorption-properties-of-metal-organic-frameworks", "title": "Towards ab-initio quality description of porous materials: Developing general Machine-Learned Potentials to simulate physical and adsorption properties of Metal-Organic Frameworks", "speaker": "Yue Yifei", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=js5pzOVcqB", "openreview_id": "js5pzOVcqB", "speaker_profile_url": "https://openreview.net/profile?id=~Yifei_Yue2" }, { "talk_id": "oral:diffusion-driven-generation-of-novel-crystalline-materials-with-target-optical-properties", "title": "Diffusion-Driven Generation of Novel Crystalline Materials with Target Optical Properties", "speaker": "TBC Ivan Kruglov", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=aQh8uinJku", "openreview_id": "aQh8uinJku", "speaker_display_name": "Ivan Kruglov; Liudmila Klimova" }, { "talk_id": "oral:atom-in-molecule-based-quantum-machine-learning-of-defect-formation-energies", "title": "Atom-in-molecule based quantum machine learning of defect formation energies", "speaker": "Alastair Price", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=3ls3YyLPzY", "openreview_id": "3ls3YyLPzY", "speaker_profile_url": "https://openreview.net/profile?id=~Alastair_James_Arthur_Price1" }, { "talk_id": "oral:ai-guided-experimental-design-of-zirconium-mops-with-the-world-avatar-for-sustainable-photocatalysis", "title": "AI-guided experimental design of zirconium MOPs with The World Avatar for sustainable photocatalysis", "speaker": "Patrick Butler", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=2O6mcYE0av", "openreview_id": "2O6mcYE0av" } ] }, { "session_title": "Self-Driving Labs for Autonomous Experimentation and Discovery", "rationale": "AI-driven LNP design for mRNA delivery, Development of a benchtop self-driving laboratory for electrocatalyst deposition and evaluation, NIMO Controller: An accessible self-driving laboratory orchestrator based on the model context protocol, AutoMEA - an automated electrolyser device for self-driving labs, Integrating Multimodal Knowledge Mining and Autonomous Experimentation for Accelerated Electrosynthesis Discovery, IvoryOS: An Interoperable Platform and Community for Self-Driving Laboratories", "section": "Self-Driving Labs", "breakout_venue_name": "Olivia", "chair_name": "Adam Gormley", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Bowen Li", "title": "AI-driven LNP design for mRNA delivery", "speaker": "Bowen Li", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "The rapid expansion of the mRNA therapeutic landscape necessitates discovery platforms that can overcome the limitations of sparse historical data and traditional trial-and-error methodologies. This talk introduces LUMI-lab, a foundation model-driven self-driving laboratory that integrates a transformer-based 3D molecular model with robotic automation to navigate vast chemical spaces. By leveraging unsupervised pretraining on over 28 million structures, the platform enables data-efficient, few-shot learning to identify high-performing ionizable lipids for lung-specific mRNA delivery. LUMI-lab autonomously synthesized and screened over 1,700 lipid nanoparticles, leading to the discovery of halogenated lipid tails as a novel structural feature that enhances endosomal escape. These efforts culminated in the identification of LUMI-6, which achieved 20.3% gene editing efficacy in lung epithelial cells in vivo. Ultimately, LUMI-lab establishes a scalable framework for the autonomous discovery of complex biomaterials, bridging the gap between artificial intelligence and translatable RNA medicine.", "speaker_profile_url": "https://www.li-bowen.com/", "section": [ "Self-Driving Labs", "AI for Biology" ] }, { "talk_id": "oral:development-of-a-benchtop-self-driving-laboratory-for-electrocatalyst-deposition-and-evaluation", "title": "Development of a benchtop self-driving laboratory for electrocatalyst deposition and evaluation", "speaker": "Shigeru Kobayashi", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=q5XQIpM6oL", "openreview_id": "q5XQIpM6oL", "speaker_profile_url": "https://openreview.net/profile?id=~Shigeru_Kobayashi1" }, { "talk_id": "oral:nimo-controller-an-accessible-self-driving-laboratory-orchestrator-based-on-the-model-context-protocol", "title": "NIMO Controller: An accessible self-driving laboratory orchestrator based on the model context protocol", "speaker": "Naruki Yoshikawa", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=kLVzE4mhCu", "openreview_id": "kLVzE4mhCu", "speaker_profile_url": "https://openreview.net/profile?id=~Naruki_Yoshikawa1" }, { "talk_id": "oral:automea-an-automated-electrolyser-device-for-self-driving-labs", "title": "AutoMEA - an automated electrolyser device for self-driving labs", "speaker": "Calvin Phan", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=h6ZJipkj8J", "openreview_id": "h6ZJipkj8J", "speaker_profile_url": "https://openreview.net/profile?id=~Calvin_Phan1" }, { "talk_id": "oral:integrating-multimodal-knowledge-mining-and-autonomous-experimentation-for-accelerated-electrosynthesis-discovery", "title": "Integrating Multimodal Knowledge Mining and Autonomous Experimentation for Accelerated Electrosynthesis Discovery", "speaker": "Han Hao", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=MQhQQsKFmj", "openreview_id": "MQhQQsKFmj", "speaker_profile_url": "https://openreview.net/profile?id=~Han_Hao1" }, { "talk_id": "oral:ivoryos-an-interoperable-platform-and-community-for-self-driving-laboratories", "title": "IvoryOS: An Interoperable Platform and Community for Self-Driving Laboratories", "speaker": "Ivory Wenyu Zhang", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=4gvtXFkDPu", "openreview_id": "4gvtXFkDPu", "speaker_profile_url": "https://openreview.net/profile?id=~Wenyu_Zhang9" } ], "chair_profile_url": "https://bme.rutgers.edu/adam-j-gormley" }, { "session_title": "Automated discovery and closed-loop optimization in chemistry and polymers", "rationale": "Learning Stochastic Polymer Dynamics at the Single-Molecule Level with High-Throughput Experiments, High-index saddle dynamics for the automated mapping of reaction routes, PlateOpt: Bayesian Optimization for Organic Catalysis in Combinatorial Well Plates, A Self-Driving Closed-Loop Workflow for Data-Efficient Kinetic Modeling and Op-timization of the Aldol Reaction, A systematic effort toward establishing an automatic end-to-end synthesis workflow for small molecules, Automation and AI-Powered Prediction in Chromatographic Separation", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Ulrich S. Schubert", "chair_profile_url": "https://www.schubert-group.uni-jena.de/en/1480/about-prof-dr-ulrich-s-schubert", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Beatrice Soh", "title": "Learning Stochastic Polymer Dynamics at the Single-Molecule Level with High-Throughput Experiments", "speaker": "Beatrice Soh", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Polymers exhibit complex behavior across multiple length scales, but traditional single-molecule experiments are slow and low-throughput, limiting statistical insight. In this talk, I will present an automated experimental platform for high-throughput observation and manipulation of individual polymer chains under nonequilibrium conditions.\n\nThe platform integrates microfluidics and real-time feedback control to trap, stretch and observe individual molecules with minimal human intervention. These experiments generate rich and large datasets that capture nonequilibrium polymer dynamics at the single-molecule level.\n\nUsing these datasets, we apply physics-guided deep learning to extract reduced coordinates and reconstruct energy landscapes, linking microscopic fluctuations to macroscopic behavior. This combination of automation and data-driven modeling enables quantitative studies of nonequilibrium polymer dynamics and lays the foundation for autonomous experimental exploration of soft matter systems.", "speaker_profile_url": "https://bwysoh.wixsite.com/sohlab", "section": [ "AI for Physics" ] }, { "talk_id": "oral:high-index-saddle-dynamics-for-the-automated-mapping-of-reaction-routes", "title": "High-index saddle dynamics for the automated mapping of reaction routes", "speaker": "Stephen Dale", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=lQ3UJfTKbG", "openreview_id": "lQ3UJfTKbG" }, { "talk_id": "oral:plateopt-bayesian-optimization-for-organic-catalysis-in-combinatorial-well-plates", "title": "PlateOpt: Bayesian Optimization for Organic Catalysis in Combinatorial Well Plates", "speaker": "Florian Boser", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=VQUKwMn4FL", "openreview_id": "VQUKwMn4FL", "speaker_profile_url": "https://openreview.net/profile?id=~Florian_Boser1" }, { "talk_id": "oral:a-self-driving-closed-loop-workflow-for-data-efficient-kinetic-modeling-and-op-timization-of-the-aldol-reaction", "title": "A Self-Driving Closed-Loop Workflow for Data-Efficient Kinetic Modeling and Op-timization of the Aldol Reaction", "speaker": "Xiao Li", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=GoEjoqvJt2", "openreview_id": "GoEjoqvJt2", "speaker_profile_url": "https://openreview.net/profile?id=~Xiao_Li45" }, { "talk_id": "oral:a-systematic-effort-toward-establishing-an-automatic-end-to-end-synthesis-workflow-for-small-molecules", "title": "A systematic effort toward establishing an automatic end-to-end synthesis workflow for small molecules", "speaker": "Chen Jie", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=Gfc48JIpVq", "openreview_id": "Gfc48JIpVq", "speaker_profile_url": "https://openreview.net/profile?id=~Jie_Chen54" }, { "talk_id": "oral:automation-and-ai-powered-prediction-in-chromatographic-separation", "title": "Automation and AI-Powered Prediction in Chromatographic Separation", "speaker": "Wendi Cai", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=5UjTJn9nr5", "openreview_id": "5UjTJn9nr5", "speaker_profile_url": "https://openreview.net/profile?id=~Wendi_CAI1" } ] }, { "session_title": "LLM Agents and Autonomous Scientific Discovery", "rationale": "Negative Space Learning: Where Survival is the Only Reward, Large Language Model Agents Enable Autonomous Design and Image Analysis of Microwell Microfluidics, Neurosymbolic Guardrails for World-Model Digital Twins: Securing AI-Driven Scientific Discovery and Autonomy, How Prompt Structural Framing and Cognitive Scaffolding Influence Performance in Generative AI Design?, Can We Automate Scientific Reasoning in Closed-Loop Experiments using Large Language Models?, Evolving collaborative research ideas with multi-agent grounding in lab-specific contexts and literature", "section": "AI Agents and LLMs for Science", "breakout_venue_name": "Morrison", "chair_name": "Tejs Vegge", "chair_profile_url": "https://www.linkedin.com/in/tejs-vegge-b1aa291/", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Jennifer Dodgson", "title": "Negative Space Learning: Where Survival is the Only Reward", "speaker": "Jennifer Dodgson", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Human knowledge occupies a small, smooth island in a vast ocean of impossible ideas. Most modern AI systems are trained to move efficiently across that island, optimizing within the boundaries of smooth and reward-dense patterns humans have already discovered.\n\nThe next generation of AI will not remain on that island. As we ask machines to explore spaces beyond current human understanding, the assumptions underlying conventional optimisation begin to break down. Outside known domains, the landscape is discontinuous, sparse, and dominated by hard constraints rather than gradual gradients: unwary explorers risk dropping off the edge of the map.\n\nThis talk introduces our work on Negative Space Learning (NSL), an alternative approach to discovery-oriented AI systems in reward-sparse regimes. We argue that superhuman AI requires open-ended learning in environments we cannot scaffold. We do not know what a solution to a 500 IQ problem looks like, and thus we cannot reward-shape our way there. Instead we must provide survival constraints and let the system find its own path. Instead of optimising directly toward a predefined notion of success, NSL systems iteratively carve away regions of failure through interaction with environments and constraint violations.", "speaker_profile_url": "https://jenniferdodgson.com/", "section": [ "ML Algorithmic Advances" ] }, { "talk_id": "oral:large-language-model-agents-enable-autonomous-design-and-image-analysis-of-microwell-microfluidics", "title": "Large Language Model Agents Enable Autonomous Design and Image Analysis of Microwell Microfluidics", "speaker": "Ngoc Duy Dinh", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=mBGPzlFuTn", "openreview_id": "mBGPzlFuTn", "speaker_profile_url": "https://openreview.net/profile?id=~Ngoc-Duy_Dinh1" }, { "talk_id": "oral:neurosymbolic-guardrails-for-world-model-digital-twins-securing-ai-driven-scientific-discovery-and-autonomy", "title": "Neurosymbolic Guardrails for World-Model Digital Twins: Securing AI-Driven Scientific Discovery and Autonomy", "speaker": "Samuel Addington", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=kyzGX7zcW3", "openreview_id": "kyzGX7zcW3", "speaker_profile_url": "https://openreview.net/profile?id=~Samuel_Addington2" }, { "talk_id": "oral:how-prompt-structural-framing-and-cognitive-scaffolding-influence-performance-in-generative-ai-design", "title": "How Prompt Structural Framing and Cognitive Scaffolding Influence Performance in Generative AI Design?", "speaker": "Yitian Huang", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=fjUnZ1kyLi", "openreview_id": "fjUnZ1kyLi", "speaker_profile_url": "https://openreview.net/profile?id=~Yitian_Huang4" }, { "talk_id": "oral:can-we-automate-scientific-reasoning-in-closed-loop-experiments-using-large-language-models", "title": "Can We Automate Scientific Reasoning in Closed-Loop Experiments using Large Language Models?", "speaker": "Mengjia Zhu", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=bWvBd55TxP", "openreview_id": "bWvBd55TxP", "speaker_profile_url": "https://openreview.net/profile?id=~Mengjia_Zhu3" }, { "talk_id": "oral:evolving-collaborative-research-ideas-with-multi-agent-grounding-in-lab-specific-contexts-and-literature", "title": "Evolving collaborative research ideas with multi-agent grounding in lab-specific contexts and literature", "speaker": "Yu Chinen", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=RDTccqN78N", "openreview_id": "RDTccqN78N", "speaker_profile_url": "https://openreview.net/profile?id=~Yu_Chinen1" } ] }, { "session_title": "Bayesian Optimization for Autonomous Materials and Process Discovery", "rationale": "Neuromorphic Systems: Towards Sustainable AI, Directing Open-Ended Evolution in Artificial Life via Temporal Multi-Scale Structural Complexity, Self-Driven Process Optimization in Pneumatic 3D Printing: From Static Ensemble Learning to Autonomous Bayesian Method, A Framework for Bayesian Optimization in Mixture Spaces, Bayesian Optimization for the Inverse Problems in Materials Science, Meta Bayesian Optimization to Discover a Problem Worth Optimizing", "section": "ML Algorithmic Advances", "breakout_venue_name": "Hullet", "chair_name": "Willi Gottstein", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Luis Camuñas-Mesa", "title": "Neuromorphic Systems: Towards Sustainable AI", "speaker": "Luis Camuñas-Mesa", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "The growing demand for computational power needed to develop and operate advanced artificial intelligence models is consuming large amounts of energy. Motivated by the efficacy, efficiency, and robustness of natural intelligence in biological information processing systems, neuromorphic engineering offers viable options for extremely low-energy cognitive computing in highly parallel, distributed architectures.\n\nThe emergence of nanoscale memristors, combined with neuromorphic hardware, has generated hope for building ultra-dense circuit architectures to perform in-memory-computing. Current hybrid CMOS-memristor technologies allow for the implementation of large-scale neuromorphic systems with dense networks of memristors in the form of crossbars, performing synaptic connections between neural layers. These systems are especially suitable for implementing online learning algorithms such as STDP (Spike-Timing-Dependent Plasticity).", "speaker_profile_url": "https://scholar.google.com/citations?user=miZTP-EAAAAJ&hl=en", "section": [ "Unconventional Computing" ] }, { "talk_id": "oral:directing-open-ended-evolution-in-artificial-life-via-temporal-multi-scale-structural-complexity", "title": "Directing Open-Ended Evolution in Artificial Life via Temporal Multi-Scale Structural Complexity", "speaker": "Andrey Ustyuzhanin", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=mlzuqP7n4C", "openreview_id": "mlzuqP7n4C" }, { "talk_id": "oral:self-driven-process-optimization-in-pneumatic-3d-printing-from-static-ensemble-learning-to-autonomous-bayesian-method", "title": "Self-Driven Process Optimization in Pneumatic 3D Printing: From Static Ensemble Learning to Autonomous Bayesian Method", "speaker": "Maxime Goulet", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=RKJAYCzH96", "openreview_id": "RKJAYCzH96", "speaker_profile_url": "https://openreview.net/profile?id=~Maxime_Goulet1" }, { "talk_id": "oral:a-framework-for-bayesian-optimization-in-mixture-spaces", "title": "A Framework for Bayesian Optimization in Mixture Spaces", "speaker": "Paola Driza", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=QwD8v255Wo", "openreview_id": "QwD8v255Wo", "speaker_profile_url": "https://openreview.net/profile?id=~Paola_Driza1" }, { "talk_id": "oral:bayesian-optimization-for-the-inverse-problems-in-materials-science", "title": "Bayesian Optimization for the Inverse Problems in Materials Science", "speaker": "Hongbin Zhang", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=M3V7YteSe2", "openreview_id": "M3V7YteSe2", "speaker_profile_url": "https://openreview.net/profile?id=~Hongbin_Zhang8" }, { "talk_id": "oral:meta-bayesian-optimization-to-discover-a-problem-worth-optimizing", "title": "Meta Bayesian Optimization to Discover a Problem Worth Optimizing", "speaker": "Yuki Takezawa", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=6Bc5wWDCef", "openreview_id": "6Bc5wWDCef" } ], "chair_profile_url": "https://openreview.net/profile?id=~Willi_Gottstein1" } ] }, { "block_name": "evening_session", "start_time": "16:45", "end_time": "18:00", "sessions": [ { "session_title": "LLMs and Autonomous Agents for Scientific Discovery and Automation", "rationale": "Reasoning in the Language of Materials, A Universal Autonomous Agent for Atomistic Simulation and Benchmarking Its Capabilities, DarkMatterFM: An Agentic Foundation Model for Multimodal Dark-Matter Inference with GPU-Accelerated Emulators, MOOSE-Chem2: Exploring LLM Limits in Fine-Grained Scientific Hypothesis Discovery via Hierarchical Search, DIGIBAT: Bridging the gap between physical automation and AI in energy research", "section": "AI Agents and LLMs for Science", "breakout_venue_name": "Morrison", "chair_name": "Ray Meng Gao", "chair_profile_url": "https://sites.google.com/view/raymgao/about", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Mohamad Moosavi", "title": "Reasoning in the Language of Materials", "speaker": "Mohamad Moosavi", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Scientific research begins with the literature: both human and AI scientists rely on prior work to synthesize ideas, identify gaps and contradictions, and formulate new hypotheses. Despite rapid advances in artificial intelligence (AI) for scientific prediction, automation, and self-driving laboratories (SDLs), this foundational step, reasoning over the scientific literature at scale, remains largely inaccessible to current systems. Existing approaches typically rely on localized retrieval of a small number of documents, limiting their ability to aggregate evidence, detect global trends, or reconcile conflicting results across thousands of studies. In this talk, I present an approach that leverages the native language of materials for literature-informed, autonomous scientific discovery. The framework treats the research literature as a first-class data modality and integrates it with machine learning and experimentation in an iterative, closed-loop system. Moving beyond isolated predictions, it enables global reasoning across prior studies, supports causal hypothesis generation, and continuously refines insights based on new evidence. I will demonstrate use cases in the design of metal–organic framework (MOF) materials for carbon capture and storage.", "speaker_profile_url": "https://chem-eng.utoronto.ca/faculty-staff/faculty-members/seyed-mohamad-moosavi/", "section": [ "AI Agents and LLMs for Science" ] }, { "talk_id": "oral:a-universal-autonomous-agent-for-atomistic-simulation-and-benchmarking-its-capabilities", "title": "A Universal Autonomous Agent for Atomistic Simulation and Benchmarking Its Capabilities", "speaker": "Fengxu Yang", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=zDszKbR8qb", "openreview_id": "zDszKbR8qb", "speaker_profile_url": "https://openreview.net/profile?id=~Fengxu_Yang1" }, { "talk_id": "oral:darkmatterfm-an-agentic-foundation-model-for-multimodal-dark-matter-inference-with-gpu-accelerated-emulators", "title": "DarkMatterFM: An Agentic Foundation Model for Multimodal Dark-Matter Inference with GPU-Accelerated Emulators", "speaker": "Ioana Zelko", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=aAeSVgoLtb", "openreview_id": "aAeSVgoLtb", "speaker_profile_url": "https://openreview.net/profile?id=~Ioana_Zelko1" }, { "talk_id": "oral:moose-chem2-exploring-llm-limits-in-fine-grained-scientific-hypothesis-discovery-via-hierarchical-search", "title": "MOOSE-Chem2: Exploring LLM Limits in Fine-Grained Scientific Hypothesis Discovery via Hierarchical Search", "speaker": "Zonglin Yang", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=3nDQdtNpXp", "openreview_id": "3nDQdtNpXp", "speaker_profile_url": "https://openreview.net/profile?id=~Zonglin_Yang1" }, { "talk_id": "oral:digibat-bridging-the-gap-between-physical-automation-and-ai-in-energy-research", "title": "DIGIBAT: Bridging the gap between physical automation and AI in energy research", "speaker": "Jingyu Feng", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=oVkRcFNZdP", "openreview_id": "oVkRcFNZdP", "speaker_profile_url": "https://openreview.net/profile?id=~Jingyu_Feng1" } ] }, { "session_title": "Autonomous materials and chemistry optimization", "rationale": "Data-Driven Materials Discovery for Rechargeable Batteries: High-Throughput Experimental Platforms and Closed-Loop Autonomous Optimization, Machine learning for in-situ composition mapping in a self-driving magnetron sputtering system, When is Bayesian Optimization Beneficial? A Critical Assessment of Optimization Strategies in High-Throughput Organic Photovoltaic Manufacturing, Constrained composite Bayesian optimisation for rational synthesis of polymeric particles, Autonomous Optimization of Perovskite Solar Cell Thin Films via Robotic Spin-Coating and Bayesian Optimization", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Keith A. Brown", "chair_profile_url": "https://kablab.org/", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Shoichi Matsuda", "title": "Data-Driven Materials Discovery for Rechargeable Batteries: High-Throughput Experimental Platforms and Closed-Loop Autonomous Optimization", "speaker": "Shoichi Matsuda", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Data-driven methodologies are transforming the discovery of functional materials by overcoming the limitations of human bias and traditional trial-and-error approaches. This is particularly critical in rechargeable battery research, where exploring the massive compositional search space of multi-component electrolytes is impractical using conventional methods. To address this challenge, researchers have developed a high-throughput, automated experimental platform featuring a closed-type 36-well multi-channel electrochemical (MCE) cell module and a non-contact dispenser. This robotic setup significantly suppresses electrolyte evaporation and enables the rapid preparation and evaluation of over 400 samples per week.\n\nTo achieve fully autonomous materials discovery, this high-throughput hardware is integrated with artificial intelligence (AI) using NIMO, a newly developed modular orchestration system. NIMO allows for flexible closed-loop workflows by connecting various AI algorithms with robotic modules. As a demonstration, the integrated system utilized Bayesian optimization to autonomously explore and optimize multi-component electrolytes for sodium-ion batteries. Through iterative closed-loop cycles, the AI successfully identified specific additive combinations that maximized the coulombic efficiency of \\ce{Na3V2(PO4)3}/hard carbon batteries. Importantly, this was achieved entirely through data-driven exploration, demonstrating the power of autonomous optimization without reliance on prior human intuition", "speaker_profile_url": "https://samurai.nims.go.jp/profiles/matsuda_shoichi", "section": [ "AI for Materials Science", "Self-Driving Labs", "AI for Chemistry" ] }, { "talk_id": "oral:machine-learning-for-in-situ-composition-mapping-in-a-self-driving-magnetron-sputtering-system", "title": "Machine learning for in-situ composition mapping in a self-driving magnetron sputtering system", "speaker": "Sanna Jarl", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=ttXdworce2", "openreview_id": "ttXdworce2", "speaker_profile_url": "https://openreview.net/profile?id=~Sanna_Jarl1" }, { "talk_id": "oral:when-is-bayesian-optimization-beneficial-a-critical-assessment-of-optimization-strategies-in-high-throughput-organic-photovoltaic-manufacturing", "title": "When is Bayesian Optimization Beneficial? A Critical Assessment of Optimization Strategies in High-Throughput Organic Photovoltaic Manufacturing", "speaker": "Matthew Osvaldo", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=ky12RSRCYc", "openreview_id": "ky12RSRCYc", "speaker_profile_url": "https://openreview.net/profile?id=~Matthew_Osvaldo1" }, { "talk_id": "oral:constrained-composite-bayesian-optimisation-for-rational-synthesis-of-polymeric-particles", "title": "Constrained composite Bayesian optimisation for rational synthesis of polymeric particles", "speaker": "Fanjin Wang", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=i5DBk6Uyl4", "openreview_id": "i5DBk6Uyl4", "speaker_profile_url": "https://openreview.net/profile?id=~Fanjin_Wang1" }, { "talk_id": "oral:autonomous-optimization-of-perovskite-solar-cell-thin-films-via-robotic-spin-coating-and-bayesian-optimization", "title": "Autonomous Optimization of Perovskite Solar Cell Thin Films via Robotic Spin-Coating and Bayesian Optimization", "speaker": "Lars Sonneveld", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=YAfT8sila5", "openreview_id": "YAfT8sila5", "speaker_profile_url": "https://openreview.net/profile?id=~Lars_Sonneveld1" } ] }, { "session_title": "Cross-domain ML for security and scientific imaging", "rationale": "Building Physical AI Systems: From Geometry and Physics to Scientific Discovery, A Three-Level Feature Selection Framework for Android Malware Detection, Exiaa: Explainable Injections for Adversarial Attack, MultiTaskDeltaNet: Change Detection-based Image Segmentation for operando ETEM with Application to Carbon Gasification Kinetics, Code and Data are not all you need for reproducibility", "section": "ML Algorithmic Advances", "breakout_venue_name": "Hullet", "chair_name": "Gianmarco Mengaldo", "chair_profile_url": "https://www.mathexlab.com/", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Ivor W. Tsang", "title": "Building Physical AI Systems: From Geometry and Physics to Scientific Discovery", "speaker": "Ivor W. Tsang", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "AI for science is transforming research across materials science, environmental science, semiconductor design, and many other scientific and engineering domains. However, many scientific challenges require more than data-driven learning alone; they demand a deep understanding of the geometry, physics, and interactions that govern real-world systems.\nThis talk introduces Physical AI Systems, built upon physical world models that integrate geometric representations, mathematical formulations, physical laws, simulation-based analysis, and data-driven intelligence. These models provide a computational understanding of environments, structures, and processes, enabling complex phenomena to be represented, explored, and validated across multiple scales, operating conditions, and design parameters.\nBy combining parametric modeling, physics-driven simulation, and AI techniques, our framework transforms geometric and physical information into actionable scientific knowledge. It supports prediction, optimization, reasoning, decision-making, and design exploration beyond traditional data-centric approaches, enhancing scientific understanding and accelerating engineering innovation in real-world applications.", "speaker_profile_url": "https://www.a-star.edu.sg/cfar/about-cfar/management/prof-ivor-tsang", "section": [ "AI for Science" ] }, { "talk_id": "oral:a-three-level-feature-selection-framework-for-android-malware-detection", "title": "A Three-Level Feature Selection Framework for Android Malware Detection", "speaker": "Abdul Kadir", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=mdZFUj6uJz", "openreview_id": "mdZFUj6uJz" }, { "talk_id": "oral:exiaa-explainable-injections-for-adversarial-attack", "title": "Exiaa: Explainable Injections for Adversarial Attack", "speaker": "Leonardo Pesce", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=PYOQRxQRgc", "openreview_id": "PYOQRxQRgc", "speaker_profile_url": "https://openreview.net/profile?id=~Leonardo_Pesce1" }, { "talk_id": "oral:multitaskdeltanet-change-detection-based-image-segmentation-for-operando-etem-with-application-to-carbon-gasification-kinetics", "title": "MultiTaskDeltaNet: Change Detection-based Image Segmentation for operando ETEM with Application to Carbon Gasification Kinetics", "speaker": "Qian Yang", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=LMY7TQ7Jhy", "openreview_id": "LMY7TQ7Jhy", "speaker_profile_url": "https://openreview.net/profile?id=~Qian_Yang6" }, { "talk_id": "oral:code-and-data-are-not-all-you-need-for-reproducibility", "title": "Code and Data are not all you need for reproducibility", "speaker": "Bohui lyu", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=3D9d8CfGU8", "openreview_id": "3D9d8CfGU8", "speaker_profile_url": "https://openreview.net/profile?id=~bohui_lyu1" } ] }, { "session_title": "AI for Materials and Molecular Discovery", "rationale": "Designing Materials That Can Be Made, Towards accelerating the discovery of efficient iridium(III) emitters using a novel database and machine learning based only on structural formulas, Data-scarce synthesis-by-design of ferroelectric Dion–Jacobson 2D hybrid organic–inorganic perovskites, SWITCH (EnterpriseSG) Talk, High-Throughput In-Device Screening of Printable Lead-Free Halide Perovskite Memristors via Machine Learning-Driven Optimization", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Victor Posligua", "chair_profile_url": "https://openreview.net/profile?id=~Victor_Posligua1", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Yousung Jung", "title": "Designing Materials That Can Be Made", "speaker": "Yousung Jung", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "The rapid growth of materials informatics has enabled computational models to propose vast numbers of candidate materials with targeted properties. However, property optimization alone is not sufficient for real-world discovery: proposed materials must also be experimentally synthesizable under feasible conditions. This talk will focus on our recent work toward synthesizability-aware materials discovery, including machine learning models that predict the likelihood of inorganic materials synthesis and identify plausible synthesis pathways. I will highlight how explainable AI can reveal the structural, chemical, and process-related factors that govern synthesis feasibility, thereby transforming prediction models into practical design tools. I will further discuss how agentic AI systems can integrate materials design, synthesis planning, feedback from experiments, and autonomous decision-making to close the loop between computation and laboratory realization.", "speaker_profile_url": "https://micc.snu.ac.kr", "section": [ "AI for Chemistry", "AI for Materials Science" ] }, { "talk_id": "oral:towards-accelerating-the-discovery-of-efficient-iridium-iii-emitters-using-a-novel-database-and-machine-learning-based-only-on-structural-formulas", "title": "Towards accelerating the discovery of efficient iridium(III) emitters using a novel database and machine learning based only on structural formulas", "speaker": "Sergei Tatarin", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=pC9RcdAOFV", "openreview_id": "pC9RcdAOFV", "speaker_profile_url": "https://openreview.net/profile?id=~Sergei_V._Tatarin1" }, { "talk_id": "oral:data-scarce-synthesis-by-design-of-ferroelectric-dionjacobson-2d-hybrid-organicinorganic-perovskites", "title": "Data-scarce synthesis-by-design of ferroelectric Dion–Jacobson 2D hybrid organic–inorganic perovskites", "speaker": "Lulu Wang", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=derQhpUPoH", "openreview_id": "derQhpUPoH" }, { "talk_id": "extra:contributed:switch-enterprisesg-talk:1", "title": "SWITCH (EnterpriseSG) Talk", "speaker": "SWITCH", "talk_type": "sponsor", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract": "" }, { "talk_id": "oral:high-throughput-in-device-screening-of-printable-lead-free-halide-perovskite-memristors-via-machine-learning-driven-optimization", "title": "High-Throughput In-Device Screening of Printable Lead-Free Halide Perovskite Memristors via Machine Learning-Driven Optimization", "speaker": "Emha Bayu Miftahullatif", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=M12LycTOdD", "openreview_id": "M12LycTOdD", "speaker_profile_url": "https://openreview.net/profile?id=~Emha_Bayu_Miftahullatif1" } ] }, { "session_title": "Scientific Agents for Discovery and Simulation", "rationale": "How Autonomous Labs & AI Are Transforming Scientific Discovery, La Agente Optima – orchestrated Bayesian optimization and active learning for accelerated in-silico compound discovery, El Agente Gráfico: Structured Execution Graph for Scientific Agents, SciAgent: Containerized Code Generation for Scientific Computing with Verification, El Agente Forjador: Task-Driven Agent Generation for Quantum Simulation", "section": "AI Agents and LLMs for Science", "breakout_venue_name": "Moor", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Santiago Miret", "title": "How Autonomous Labs & AI Are Transforming Scientific Discovery", "speaker": "Santiago Miret", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "For most of its history, the scientific method has followed a familiar rhythm: hypothesize, design, experiment, learn, repeat. Recent advances in AI, machine learning, and automated systems enable us to ask deeper questions about how science changes when all of these technologies come together. In this session, Santiago Miret will introduce Lila Sciences and its vision for building an Operating System for Science that leverages a platform combining advanced AI reasoning models with autonomous robot-centric laboratories (AI Science Factories™) to drive new scientific discovery. Santiago will share concrete examples of how this closed-loop paradigm is accelerating discovery across materials science, chemistry, and the life sciences, including non-platinum catalysts for green hydrogen production to novel sorbents for carbon capture.", "speaker_profile_url": "https://linkedin.com/in/santiago-miret", "section": [ "AI for Materials Science", "AI Agents and LLMs for Science" ] }, { "talk_id": "oral:la-agente-optima-orchestrated-bayesian-optimization-and-active-learning-for-accelerated-in-silico-compound-discovery", "title": "La Agente Optima – orchestrated Bayesian optimization and active learning for accelerated in-silico compound discovery", "speaker": "Marcel Mueller", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=fVnYXtbkou", "openreview_id": "fVnYXtbkou", "speaker_profile_url": "https://openreview.net/profile?id=~Marcel_Müller2" }, { "talk_id": "oral:el-agente-grafico-structured-execution-graph-for-scientific-agents", "title": "El Agente Gráfico: Structured Execution Graph for Scientific Agents", "speaker": "Jiaru Bai", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=YShauPAyTw", "openreview_id": "YShauPAyTw", "speaker_profile_url": "https://openreview.net/profile?id=~Jiaru_Bai1" }, { "talk_id": "oral:sciagent-containerized-code-generation-for-scientific-computing-with-verification", "title": "SciAgent: Containerized Code Generation for Scientific Computing with Verification", "speaker": "Shruti Badhwar", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=FzcjjKIoqt", "openreview_id": "FzcjjKIoqt", "speaker_profile_url": "https://openreview.net/profile?id=~Shruti_Badhwar1" }, { "talk_id": "oral:el-agente-forjador-task-driven-agent-generation-for-quantum-simulation", "title": "El Agente Forjador: Task-Driven Agent Generation for Quantum Simulation", "speaker": "Zijian Zhang", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=7aXeu0hHo5", "openreview_id": "7aXeu0hHo5", "speaker_profile_url": "https://openreview.net/profile?id=~Zijian_Zhang12" } ], "chair_name": "Andrey E Ustyuzhanin", "chair_profile_url": "https://openreview.net/profile?id=~Andrey_E_Ustyuzhanin1" } ] } ], "breaks": [ { "name": "Tea Break", "start_time": "10:30", "end_time": "11:00", "sponsor": "SEA Garena" }, { "name": "Break", "start_time": "12:00", "end_time": "12:15" }, { "name": "Lunch", "start_time": "13:15", "end_time": "14:45", "sponsor": "JEOL Asia Pte Ltd" }, { "name": "Tea Break", "start_time": "16:15", "end_time": "16:45" } ], "midday_chair_profile_url": "https://openreview.net/profile?id=~Kostya_S._Novoselov1" }, { "date": "2026-06-17", "plenary_talks": [ { "talk_id": "invited:Eun-Ah Kim", "title": "Learning Quantum Matter from Data: Data Centric AI for Scientific Discovery", "speaker": "Eun-Ah Kim", "talk_type": "plenary", "length": 1, "start_time": "09:00", "end_time": "09:30", "abstract": "Artificial intelligence is beginning to reshape the scientific method itself. It has the potential to discover new representations, new descriptors, and new pathways to understanding in domains where first-principles approaches alone remain incomplete. Quantum materials are a particularly rich frontier: they exhibit some of the richest emergent phenomena in nature, yet the relevant signatures are often buried in high-dimensional experiments and fragmented across heterogeneous materials data. In this talk, I will present a vision for learning quantum matter from data through two case studies: X-TEC, an unsupervised framework for extracting emergent order from evolving diffraction data[1,2], and GPTc, an interpretable, structure-aware framework for predicting superconducting transition temperatures and guiding materials discovery [3]. These examples suggest a broader future for AI in physics: not as a black box, but as an instrument of scientific insight. When grounded in physical structure and coupled to experiment, AI can help reveal the organizing principles of complex matter and accelerate discovery. \n[1] Venderley at al, PNAS 119, e2109665119 (2022).\n[2] Mallayya et al, Nature Physics 20, 822 (2024).\n[3] Lesser et al, arXiv 2510.07373 (2025).", "speaker_profile_url": "https://en.wikipedia.org/wiki/Eun-Ah_Kim", "section": [ "AI for Physics" ] }, { "talk_id": "invited:Giacomo Indiveri", "title": "Bridging natural and artificial intelligence with mixed-signal neuromorphic circuits", "speaker": "Giacomo Indiveri", "talk_type": "plenary", "length": 1, "start_time": "11:00", "end_time": "11:30", "abstract": "While machine learning has made tremendous progress in recent years, there is still a large gap between artificial and natural intelligence.\nClosing this gap requires combining fundamental research in neuroscience with mathematics, physics, and engineering to understand the principles of neural computation and cognition.\nMixed-signal subthreshold analog and asynchronous digital electronic integrated circuits offer an additional means of exploring neural computation, by providing a computational substrate that shares many similarities with the one of biological brains.\nIn this subthreshold region of operation, transistor channels employ the same physics of carrier transport (diffusion) as the proteic channels of real neurons.\nThus, complex neuromorphic circuits and networks built following this approach share many similarities with real synapses, neurons, and cortical neural circuits.\nIn this presentation I will demonstrate how to build neuromorphic processors that use the physics of their computational substrate to directly emulate the physics of biological neural processes in real-time. I will demonstrate how to build complex recurrent electronic neural circuits with dynamics and response properties strikingly similar to those measured in real neural networks. I will argue that these systems can be used to complement numerical simulations in basic research and real-world applications.", "speaker_profile_url": "https://www.ini.uzh.ch/", "section": [ "AI for Biology", "Unconventional Computing", "AI for Medicine and Healthcare" ] }, { "talk_id": "invited:Jun Jiang", "title": "Building a Global Infrastructure for AI-Driven Innovation", "speaker": "Jun Jiang", "talk_type": "plenary", "length": 1, "start_time": "12:00", "end_time": "12:30", "abstract": "AI-driven scientific innovation is often blocked by scarce, fragmented, and biased data. To solve this, a global infrastructure of cloud-connected, autonomous laboratories has been proposed, generating high-quality, reproducible datasets via robotic experiments. This vision is structured into five hierarchical levels: G1 (Process Automation), G2 (Theory-Experiment Iterative Loop), G3 (Large Model-Driven), G4 (Multi-Platform, Multi-Task), G5 (Autonomous Scientific Discovery).\nOur G4-level large-model-driven autonomous platform coordinates domain-specific small AI models and robotic experiments, enabling intelligent scheduling and real-time fine-tuning of pre-trained models based on experimental data, thereby creating a dynamic, synergistic feedback loop. Major breakthroughs in novel material creation (e.g., catalysts, polymers, COFs, and proteins) have compressed discovery cycles from over a century to mere months.\nThe global infrastructure transforms isolated scientific efforts into a collaborative and efficient exploration engine. It democratizes access to high-quality data, breaks down geographical and institutional barriers to innovation, ultimately accelerates industrial-scale scientific and technological advancements.", "speaker_profile_url": "https://faculty.ustc.edu.cn/jiangjun1/en/index.htm", "section": [ "AI for Chemistry", "AI Agents and LLMs for Science", "Self-Driving Labs" ] } ], "keynote_talks": [ { "talk_id": "extra:keynote:editors-panel-scientific-publishing-in-the-ai-era:1", "title": "Editors Panel - Scientific Publishing in the AI Era", "speaker": "Yanfei Zhu (Nature); Anna Rulka (Royal Society of Chemistry); Steve Cranford (Cell Press); Steinn Sigurdsson (arXiv); Nancy F. Chen (NeurIPS); Moderator: Alán Aspuru-Guzik (Acceleration Consortium / NVIDIA)", "talk_type": "extra_keynote", "length": 4, "start_time": "09:30", "end_time": "10:30", "abstract": "Artificial intelligence is changing not only how research is conducted, but how it is written, reviewed, published, and discovered. In this moderated, audience-driven discussion, panelists will explore what these shifts mean for scientific publishing, touching on peer review, accessibility, integrity, and other questions facing the field. Audience questions will guide a conversation on the opportunities, challenges, and uncertainties ahead." }, { "talk_id": "invited:Ngiam Kee Yuan", "title": "CASCADE AI - An Agentic AI Variant-to-Disease Mechanism Discovery", "speaker": "Ngiam Kee Yuan", "talk_type": "featured", "length": 1, "start_time": "11:30", "end_time": "11:45", "abstract": "Understanding the mechanistic basis of genetic variants in human disease remains a fundamental challenge in genomics. Evidence linking variants to disease is distributed across numerous heterogeneous sources — including large-scale consortia such as ENCODE, GTEx, and the Roadmap Epigenomics Project — yet no existing platform integrates these resources to resolve variant-to-disease relationships in a comprehensive, end-to-end mechanistic manner. Current tools, such as Open Targets, primarily aggregate structured datasets like GWAS and eQTL evidence, constraining both the coverage and depth of biological insight they can provide.\nCASCADE AI addresses this critical gap by generating mechanistic variant-disease hypotheses from a unified, multi-source database encompassing 27 or more curated datasets. This represents one of the most densely integrated genomic evidence resources assembled to characterise variant mechanisms, spanning functional, regulatory, and association data across diverse biological contexts. By consolidating this breadth of evidence, CASCADE AI moves beyond simple data aggregation to support the discovery of novel variant mechanisms and potential therapeutic targets that would otherwise remain obscured within siloed repositories.\nA key distinguishing feature of CASCADE AI is its use of agentic AI, enabling dynamic, interactive querying that adapts to user-defined research questions. Results are delivered as interpretable natural language insights, making complex multi-layered genomic evidence accessible to both computational and clinical researchers. Together, these capabilities position CASCADE AI as a transformative tool for precision medicine, accelerating the translation of genetic association data into actionable mechanistic and therapeutic understanding.", "speaker_profile_url": "https://medicine.nus.edu.sg/dbmi/about-us/faculty/ngiam-kee-yuan/", "section": [ "AI for Medicine and Healthcare", "AI for Biology", "AI Agents and LLMs for Science" ] }, { "talk_id": "invited:Aruhan Rui Shi", "title": "Macroeconomic Modeling and Forecasting with AI Tools", "speaker": "Aruhan Rui Shi", "talk_type": "featured", "length": 1, "start_time": "11:45", "end_time": "12:00", "abstract": "This talk presents applications of AI tools in macroeconomic modeling and forecasting, with a focus on using AI agents to study expectation formation in a changing environment. I also briefly discuss related work using LLM-based methods to forecast the federal funds rate from macroeconomic and policy information.", "speaker_profile_url": "https://aruhanruishi.com/", "section": [ "AI for Society", "AI for Finance" ] } ], "early_morning_chair_name": "Yong Tao Tan", "late_morning_chair_name": "Jennifer Dodgson", "late_morning_chair_profile_url": "https://jenniferdodgson.com/", "time_blocks": [ { "block_name": "afternoon_session", "start_time": "14:45", "end_time": "16:15", "sessions": [ { "session_title": "Autonomous AI for Materials Discovery and Synthesis", "rationale": "Machine learning for synthesis of real materials, Accelerating materials innovation through automated theoretical-experimental iterations empowered by AI-Chemist, Accelerating Nanocarbon Dispersion Research via Machine Learning and Automated Experimentation, Automated Bulk Intermetallic Synthesis via Orchestrated Heterogeneous Laboratory Machines, Exploration of Ternary Thin-Film Lithium Solid Electrolyte Composites Using the Digital Laboratory for Enhanced Lithium-Ion Conductivity, Closed Loop Inorganic Material Discovery with Design-Test-Make-Analyze Paradigm", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Yeong Wai Yee", "chair_profile_url": "https://www.yeongresearch.com/", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Antonio Helio Castro Neto", "title": "Machine learning for synthesis of real materials", "speaker": "Antonio Helio Castro Neto", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "This talk presents how we use machine learning to radically accelerate the discovery of next‑generation construction materials within URBAX, a green cement and kaolin-based platform. By combining experimental data, physics-informed features, and modern ML models, we explore previously uncharted regions of compositional and processing space. This allows us to design binders that reach international construction strength and durability standards in only 5 days, instead of the traditional 28-day qualification window. Beyond speed, the approach reveals non-intuitive formulations and trade-offs that link sustainability, performance, and cost, illustrating how AI can transform the way we engineer materials for the built environment and, more broadly, how data-driven discovery can reshape resource-intensive industries.", "section": [ "AI for Materials Science" ], "speaker_profile_url": "https://www.linkedin.com/in/antonio-h-castro-neto-ba8187ab?utm_source=share_via&utm_content=profile&utm_medium=member_ios" }, { "talk_id": "oral:accelerating-materials-innovation-through-automated-theoretical-experimental-iterations-empowered-by-ai-chemist", "title": "Accelerating materials innovation through automated theoretical-experimental iterations empowered by AI-Chemist", "speaker": "Zhuoying Zhu", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=xwxF2IFqOC", "openreview_id": "xwxF2IFqOC", "speaker_profile_url": "https://openreview.net/profile?id=~Zhuoying_Zhu1" }, { "talk_id": "oral:accelerating-nanocarbon-dispersion-research-via-machine-learning-and-automated-experimentation", "title": "Accelerating Nanocarbon Dispersion Research via Machine Learning and Automated Experimentation", "speaker": "Hirokuni Jintoku", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=Vxj12TtqSu", "openreview_id": "Vxj12TtqSu", "speaker_profile_url": "https://openreview.net/profile?id=~Hirokuni_Jintoku1" }, { "talk_id": "oral:automated-bulk-intermetallic-synthesis-via-orchestrated-heterogeneous-laboratory-machines", "title": "Automated Bulk Intermetallic Synthesis via Orchestrated Heterogeneous Laboratory Machines", "speaker": "Kensei Terashima", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=K86x6XyxiQ", "openreview_id": "K86x6XyxiQ", "speaker_profile_url": "https://openreview.net/profile?id=~Kensei_Terashima1" }, { "talk_id": "oral:exploration-of-ternary-thin-film-lithium-solid-electrolyte-composites-using-the-digital-laboratory-for-enhanced-lithium-ion-conductivity", "title": "Exploration of Ternary Thin-Film Lithium Solid Electrolyte Composites Using the Digital Laboratory for Enhanced Lithium-Ion Conductivity", "speaker": "Kazunori Nishio", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=7P0VDZiJaG", "openreview_id": "7P0VDZiJaG" }, { "talk_id": "oral:closed-loop-inorganic-material-discovery-with-design-test-make-analyze-paradigm", "title": "Closed Loop Inorganic Material Discovery with Design-Test-Make-Analyze Paradigm", "speaker": "Haiwen Dai", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=40tdIqAzqu", "openreview_id": "40tdIqAzqu", "speaker_profile_url": "https://openreview.net/profile?id=~Haiwen_Dai1" } ] }, { "session_title": "Self-Driving Labs for Chemistry and Materials Optimization", "rationale": "Self-driving lab for viscous nanoemulsions, Toward Generalizable, Data-Efficient Self-Driving Laboratories for Organic Materials, An Integrated Platform for In Situ Electroanalytical–Driven Reaction Optimization, Designing of Microfluidic Concentration Generator Module for Self-Driving Fluid Mixing System, Resource-efficient Bayesian optimization for self-calibrating liquid handling, ACHT-World: Causal World Models for Closed-Loop Self-Driving Laboratories", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Yang Cao", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Nasim Abdollahi", "title": "Self-driving lab for viscous nanoemulsions", "speaker": "Nasim Abdollahi", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Nanoemulsions are a class of biphasic colloidal materials comprised of 20-500-nm oil droplets dispersed in a continuous aqueous phase. These materials are capable of compatibilizing materials with dissimilar hydrophilicities, leading to their widespread use across many industries, including personal care products, pharmaceuticals, and agriculture. Synthesizing nanoemulsions involves balancing ingredients (often 5-10 at a time) and processing parameters (e.g., temperature, mixing conditions) to achieve small and stable droplets. Higher order interactions dominate this process, causing many formulators to rely on heuristics when developing new formulations. Moreover, a lack of synthesis standardization complicates the ability to learn from existing data or replicate any observed trends.\nWe have thus combined a chemistry-aware machine learning model (brain) and an automated synthesis platform (body) to produce a first-in-class self-driving lab (SDL) for nanoemulsions.\nThe brain is a high-performing and flexible ML model with a multi-class classification head for emulsification success. After demonstrating competitive performance with baseline models, it was deployed within a custom in-house graphical user interface (GUI) supporting the end-to-end machine learning workflow and database management, enabling formulators to simulate experiments, interpret model predictions and query results.\nThe body is an automated synthesis platform with integrated dispensing, heating and mixing unit operations. Each unit operation is a standalone module individually controlled by a central orchestrator, enabling modularity in platform design. Custom mounts and interfaces are used to improve throughput and versatility. Initial tests confirm successful end-to-end automated synthesis that outperforms manual methods in both speed and consistency.\nFuture work will integrate the brain and body for fully autonomous operation and accelerated formulation discovery.", "speaker_profile_url": "https://nasimabdollahi.github.io/Research.html", "section": [ "Self-Driving Labs" ] }, { "talk_id": "oral:toward-generalizable-data-efficient-self-driving-laboratories-for-organic-materials", "title": "Toward Generalizable, Data-Efficient Self-Driving Laboratories for Organic Materials", "speaker": "Martin Seifrid", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=um2t28K599", "openreview_id": "um2t28K599", "speaker_profile_url": "https://openreview.net/profile?id=~Martin_Seifrid1" }, { "talk_id": "oral:an-integrated-platform-for-in-situ-electroanalyticaldriven-reaction-optimization", "title": "An Integrated Platform for In Situ Electroanalytical–Driven Reaction Optimization", "speaker": "Timothy McClure", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=nRE3mG55xU", "openreview_id": "nRE3mG55xU", "speaker_profile_url": "https://openreview.net/profile?id=~Timothy_J._McClure1" }, { "talk_id": "oral:designing-of-microfluidic-concentration-generator-module-for-self-driving-fluid-mixing-system", "title": "Designing of Microfluidic Concentration Generator Module for Self-Driving Fluid Mixing System", "speaker": "Jeongwook Lim", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=YGQonaRHoI", "openreview_id": "YGQonaRHoI", "speaker_profile_url": "https://openreview.net/profile?id=~Jeongwook_Lim1" }, { "talk_id": "oral:resource-efficient-bayesian-optimization-for-self-calibrating-liquid-handling", "title": "Resource-efficient Bayesian optimization for self-calibrating liquid handling", "speaker": "Owen Melville", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=DQCOJbW2eY", "openreview_id": "DQCOJbW2eY", "speaker_profile_url": "https://openreview.net/profile?id=~Owen_Alfred_Melville1" }, { "talk_id": "oral:acht-world-causal-world-models-for-closed-loop-self-driving-laboratories", "title": "ACHT-World: Causal World Models for Closed-Loop Self-Driving Laboratories", "speaker": "David Scott Lewis", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=3e5S1HNvea", "openreview_id": "3e5S1HNvea", "speaker_profile_url": "https://openreview.net/profile?id=~David_Scott_Lewis1" } ], "chair_profile_url": "https://openreview.net/profile?id=~Yang_Cao38" }, { "session_title": "AI for Healthcare, Therapeutics, and Clinical Decision Support", "rationale": "Accelerating Biological Discovery Through AI, Robotics, and Human Organ Mimicry, Inferring Oocyte Cytoplasmic Material Properties from Cytoplasmic Streaming Movies Using Physics-Informed Neural Networks, Inferring the hidden and long-range dengue transmission routes in Singapore, Event driven neural network on a mixed signal neuromorphic processor for detecting EEG based epileptic seizure, Conceptualising Case Formulation as a Neurosymbolic AI Framework for Mental Health, The Cognitive Clinical OS: Architecting Asynchronous Agentic Reasoning for Real-Time Decision Support", "section": "AI for Medicine and Healthcare", "breakout_venue_name": "Morrison", "chair_name": "Luis Camuñas-Mesa", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Yimu Zhao", "title": "Accelerating Biological Discovery Through AI, Robotics, and Human Organ Mimicry", "speaker": "Yimu Zhao", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Artificial intelligence has advanced rapidly, yet its impact in biomedicine remains limited by the quality and physiological relevance of the experimental models used to generate training data. At the Self-Driving Lab for Human Organ Mimicry (SDL6), we are developing human tissue models that better represent human biology and support autonomous experimentation. Our work combines label-free quality control of induced pluripotent stem cell (iPSC) maintenance and expansion, automated fabrication and functional evaluation of engineered cardiac microtissues, and robotic generation of vascularized multicellular co-culture systems spanning multiple organ types. Together, these capabilities enable reproducible production, long-term culture, and quantitative characterization of complex human tissue models. By integrating advanced biological models with automation and data-driven analysis, SDL6 aims to improve experimental reproducibility, generate content-rich biological datasets, and accelerate the development of predictive platforms for drug discovery, disease modeling, and regenerative medicine.", "speaker_profile_url": "https://acceleration.utoronto.ca/people/yimu-zhao", "section": [ "Self-Driving Labs" ] }, { "talk_id": "oral:inferring-oocyte-cytoplasmic-material-properties-from-cytoplasmic-streaming-movies-using-physics-informed-neural-networks", "title": "Inferring Oocyte Cytoplasmic Material Properties from Cytoplasmic Streaming Movies Using Physics-Informed Neural Networks", "speaker": "Alokendra Ghosh", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=ks0tFnz4M7", "openreview_id": "ks0tFnz4M7" }, { "talk_id": "oral:inferring-the-hidden-and-long-range-dengue-transmission-routes-in-singapore", "title": "Inferring the hidden and long-range dengue transmission routes in Singapore", "speaker": "Zhen Yuan Yeo", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=v1WFU9Ft0B", "openreview_id": "v1WFU9Ft0B", "speaker_profile_url": "https://openreview.net/profile?id=~Yeo_Zhen_Yuan1" }, { "talk_id": "oral:event-driven-neural-network-on-a-mixed-signal-neuromorphic-processor-for-detecting-eeg-based-epileptic-seizure", "title": "Event driven neural network on a mixed signal neuromorphic processor for detecting EEG based epileptic seizure", "speaker": "Giacomo Indiveri", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=Vm32eFryQ4", "openreview_id": "Vm32eFryQ4" }, { "talk_id": "oral:conceptualising-case-formulation-as-a-neurosymbolic-ai-framework-for-mental-health", "title": "Conceptualising Case Formulation as a Neurosymbolic AI Framework for Mental Health", "speaker": "Poorva Pandya", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=N402RIxSbV", "openreview_id": "N402RIxSbV" }, { "talk_id": "oral:the-cognitive-clinical-os-architecting-asynchronous-agentic-reasoning-for-real-time-decision-support", "title": "The Cognitive Clinical OS: Architecting Asynchronous Agentic Reasoning for Real-Time Decision Support", "speaker": "Malik Saif", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=Gm2ldRNgHD", "openreview_id": "Gm2ldRNgHD", "speaker_profile_url": "https://openreview.net/profile?id=~Malik_Muhammad_Saif1" } ], "chair_profile_url": "https://scholar.google.com/citations?user=miZTP-EAAAAJ&hl=en" }, { "session_title": "Neural Dynamics and Theoretical Deep Learning", "rationale": "Thermodynamical Analogies in Deep Learning, Towards Critical Branching Mechanism in Recurrent Neural Networks, Order-chaos transition in deep neural network and its application to the training process, Scalable learning of macroscopic stochastic dynamics, Learning non-equilibrium mesoscopic dynamics with Onsager principle, Learning Permutation-invariant Macroscopic Dynamics", "section": "ML Algorithmic Advances", "breakout_venue_name": "Hullet", "chair_name": "Carlo Vittorio Cannistraci", "chair_profile_url": "https://brain.tsinghua.edu.cn/en/info/1010/1003.htm", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Dmitry Vetrov", "title": "Thermodynamical Analogies in Deep Learning", "speaker": "Dmitry Vetrov", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "The stochastic optimization of a loss function during the training of deep neural networks shares many similarities with classical thermodynamical systems. By analysing stochastic differential equations that describe the evolution of (scale-invariant) neural network during training we derive the characteristics of its stationary state. Surprisingly it becomes very similar to ideal gas law. Following this similarity one may define analogues of temperature, pressure, and volume for neural networks. Using those analogies we establish various thermodynamic potentials such as Gibbs and Helmholtz free energies and show that they are minimized during training under popular training protocols.", "speaker_profile_url": "https://constructor.university/faculty-member/dmitry-vetrov", "section": [ "ML Algorithmic Advances" ] }, { "talk_id": "oral:towards-critical-branching-mechanism-in-recurrent-neural-networks", "title": "Towards Critical Branching Mechanism in Recurrent Neural Networks", "speaker": "Feixiang Ren", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=ywmENyEGXM", "openreview_id": "ywmENyEGXM", "speaker_profile_url": "https://openreview.net/profile?id=~Feixiang_Ren1" }, { "talk_id": "oral:order-chaos-transition-in-deep-neural-network-and-its-application-to-the-training-process", "title": "Order-chaos transition in deep neural network and its application to the training process", "speaker": "Ling Feng", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=jz4VfXJPMS", "openreview_id": "jz4VfXJPMS", "speaker_profile_url": "https://openreview.net/profile?id=~Ling_Feng2" }, { "talk_id": "oral:scalable-learning-of-macroscopic-stochastic-dynamics", "title": "Scalable learning of macroscopic stochastic dynamics", "speaker": "Mengyi Chen", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=Kn7noYh2Gr", "openreview_id": "Kn7noYh2Gr", "speaker_profile_url": "https://openreview.net/profile?id=~Mengyi_Chen1" }, { "talk_id": "oral:learning-non-equilibrium-mesoscopic-dynamics-with-onsager-principle", "title": "Learning non-equilibrium mesoscopic dynamics with Onsager principle", "speaker": "Zhuoyuan Li", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=KCQFEZt77H", "openreview_id": "KCQFEZt77H", "speaker_profile_url": "https://openreview.net/profile?id=~Zhuoyuan_Li1" }, { "talk_id": "oral:learning-permutation-invariant-macroscopic-dynamics", "title": "Learning Permutation-invariant Macroscopic Dynamics", "speaker": "Zhichao Han", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=8kzj2c6uU5", "openreview_id": "8kzj2c6uU5", "speaker_profile_url": "https://openreview.net/profile?id=~Zhichao_Han4" } ] }, { "session_title": "AI-driven electrochemistry and catalyst discovery", "rationale": "Machine Learning Accelerated Simulations of Electrochemical Interfaces, Active Learning Interatomic Potentials-Enhanced Molecular Dynamics for Grain Boundary Engineering in Antiperovskite Solid Electrolytes, Autonomous Discovery of High-performance Ni–Mo Electrocatalysts for Green Hydrogen Production, Large Language Model Assisted Optimisation of Photocatalytic Hydrogen Production, A reinforcement learning approach to generate equivalent circuit models for Electrochemical Impedance Spectroscopy, Benchmarking Foundation Potentials against Quantum Chemistry Methods for Predicting Molecular Redox Potentials", "section": "AI for Chemistry", "breakout_venue_name": "Moor", "chair_name": "Yizhou Zhu", "chair_profile_url": "https://en.westlake.edu.cn/faculty/yizhou-zhu.html", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Pengfei Ou", "title": "Machine Learning Accelerated Simulations of Electrochemical Interfaces", "speaker": "Pengfei Ou", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "The activity–stability balance of the oxygen evolution reaction (OER) electrocatalysts is dictated by how interfacial oxidation states gate both O–O\nbond formation and metal dissolution, yet these coupled events are rarely accessible to atomistic simulations at electrified solid–liquid interfaces. Here we combine a Ru–O–H machine-learned interatomic potential with enhanced-sampling molecular dynamics to resolve oxygen evolution and Ru detachment on RuO (110) in explicit water across distinct Ru oxidation environments. We find that moderately over-oxidized, *O-ligated Ru centers promote an oxide-path mechanism (OPM) in which adjacent *O species couple with a low free-energy barrier, while a solvent-structured “water gap” suppresses nucleophilic attack and disfavors *OOH formation via the adsorbate evolution mechanism (AEM). As the surface is reduced toward Ru4+-like states and adjacent *O motifs are\ndepleted, the O–O coupling barrier rises, and the dominant pathway shifts back to AEM. In parallel, Ru dissolution proceeds through an oxidation-assisted route: additional water ligation stabilizes higher-valent Ru (Ru5+-like) and lowers the detachment barrier by ~1 eV relative to direct dissolution, rendering coordinatively unsaturated sites more vulnerable than bridge-bonded Ru. Guided by this mechanistic map, we develop an electrochemical pre-activation protocol that biases RuO2 toward an optimally over-oxidized interfacial state, enhancing OER performance while limiting Ru loss. These results establish Ru oxidation state as a unifying descriptor for controlling competing OER and degradation pathways at electrochemical interfaces.", "speaker_profile_url": "https://www.pengfeiou.com/", "section": [ "AI for Chemistry" ] }, { "talk_id": "oral:active-learning-interatomic-potentials-enhanced-molecular-dynamics-for-grain-boundary-engineering-in-antiperovskite-solid-electrolytes", "title": "Active Learning Interatomic Potentials-Enhanced Molecular Dynamics for Grain Boundary Engineering in Antiperovskite Solid Electrolytes", "speaker": "Haobo Li", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=Ojl0XhhMY1", "openreview_id": "Ojl0XhhMY1", "speaker_profile_url": "https://openreview.net/profile?id=~Haobo_Li5" }, { "talk_id": "oral:autonomous-discovery-of-high-performance-nimo-electrocatalysts-for-green-hydrogen-production", "title": "Autonomous Discovery of High-performance Ni–Mo Electrocatalysts for Green Hydrogen Production", "speaker": "Paolo Vincenzo Freiesleben de Blasio", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=I2kU55fayT", "openreview_id": "I2kU55fayT", "speaker_profile_url": "https://openreview.net/profile?id=~Paolo_Vincenzo_Freiesleben_de_Blasio1" }, { "talk_id": "oral:large-language-model-assisted-optimisation-of-photocatalytic-hydrogen-production", "title": "Large Language Model Assisted Optimisation of Photocatalytic Hydrogen Production", "speaker": "Qi Jie Yeow", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=5PtEu7mrcS", "openreview_id": "5PtEu7mrcS", "speaker_profile_url": "https://openreview.net/profile?id=~Qi_Jie_Yeow1" }, { "talk_id": "oral:a-reinforcement-learning-approach-to-generate-equivalent-circuit-models-for-electrochemical-impedance-spectroscopy", "title": "A reinforcement learning approach to generate equivalent circuit models for Electrochemical Impedance Spectroscopy", "speaker": "Yonatan Kurniawan", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=3UL9IqU5lv", "openreview_id": "3UL9IqU5lv", "speaker_profile_url": "https://openreview.net/profile?id=~Yonatan_Kurniawan1" }, { "talk_id": "oral:benchmarking-foundation-potentials-against-quantum-chemistry-methods-for-predicting-molecular-redox-potentials", "title": "Benchmarking Foundation Potentials against Quantum Chemistry Methods for Predicting Molecular Redox Potentials", "speaker": "Yicheng Chen", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=189rTnx4Yv", "openreview_id": "189rTnx4Yv", "speaker_profile_url": "https://openreview.net/profile?id=~Yicheng_Chen3" } ] } ] }, { "block_name": "evening_session", "start_time": "16:45", "end_time": "18:00", "sessions": [ { "session_title": "AI for Materials Discovery and Property Prediction", "rationale": "Machine Learning for 3D Printed Soft Robotics and Intelligent Systems, A Multimodal Conditional JEPA for Composite Materials, Test-Time Self-Evolution in Multi-Agent Systems for Materials Discovery, Data-Driven Property Prediction for Memristor Resistive Switching Layers, EMOS: The Unified AI Platform for Electronic Materials Discovery", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Seunghwa Ryu", "chair_profile_url": "https://sites.google.com/site/seunghwalab/home?authuser=0", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Yeong Wai Yee", "title": "Machine Learning for 3D Printed Soft Robotics and Intelligent Systems", "speaker": "Yeong Wai Yee", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "There has been widespread adoption of machine learning (ML) technologies to discover new insights of the complex relationships among diverse parameters in various additive manufacturing (AM) techniques. The collaborative synergy between ML and AM holds the potential to achieve interesting function and drive new generation design of AM‐printed parts. In our work, we demonstrated such synergy for 3D printed soft robotics, with optimization and control enabled by ML. We will first present the overall potential of ML in materials and processes of AM, before presenting 2 interesting examples of soft robotics. We developed a 3D printing‐enabled artificially innervated smart soft gripper with variable joint stiffness where ML was used for joint angle prediction. Furthering the research for intelligent system, we recently reported a soft continuum robot that integrates a conductive polymer composite directly into a node-based lattice structure for intrinsic sensing, coupled with a neural network for near-real-time shape reconstruction. This approach enables near-real-time proprioception without compromising flexibility, with potential applications in biomedical manipulation and remote inspection.", "speaker_profile_url": "https://www.yeongresearch.com/", "section": [ "AI for Materials Science" ] }, { "talk_id": "oral:a-multimodal-conditional-jepa-for-composite-materials", "title": "A Multimodal Conditional JEPA for Composite Materials", "speaker": "Hangwei Qian", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=uYHIlkfZuP", "openreview_id": "uYHIlkfZuP", "speaker_profile_url": "https://openreview.net/profile?id=~Hangwei_Qian1" }, { "talk_id": "oral:test-time-self-evolution-in-multi-agent-systems-for-materials-discovery", "title": "Test-Time Self-Evolution in Multi-Agent Systems for Materials Discovery", "speaker": "Bo Hu", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=qqT5zb5dnq", "openreview_id": "qqT5zb5dnq", "speaker_profile_url": "https://openreview.net/profile?id=~Bo_Hu13" }, { "talk_id": "oral:data-driven-property-prediction-for-memristor-resistive-switching-layers", "title": "Data-Driven Property Prediction for Memristor Resistive Switching Layers", "speaker": "Ben Rowlinson", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=eO1NUxjtdJ", "openreview_id": "eO1NUxjtdJ", "speaker_profile_url": "https://openreview.net/profile?id=~Ben_D._Rowlinson1" }, { "talk_id": "oral:emos-the-unified-ai-platform-for-electronic-materials-discovery", "title": "EMOS: The Unified AI Platform for Electronic Materials Discovery", "speaker": "Atish Dixit", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=crnkPDmUAd", "openreview_id": "crnkPDmUAd", "speaker_profile_url": "https://openreview.net/profile?id=~Atish_Dixit1" } ] }, { "session_title": "Closed-Loop Autonomous Labs for Chemistry and Materials Discovery", "rationale": "Closed-Loop autonomous discovery of functional membranes and 2D Materials for resource recovery and energy applications, Closed-loop Optimization of Mono-functionalization via Suzuki-Miyaura Reaction, Flow Chemistry as a Platform for Experimental Multi-objective Optimization of Heterogeneous Polymer Synthesis, NIMO: Universal Middleware for Closed-Loop Materials Exploration, MCP-Enabled LLM Agents for Closed-Loop Optimization in Real-Time Physical Experiments", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Shoichi Matsuda", "chair_profile_url": "https://samurai.nims.go.jp/profiles/matsuda_shoichi", "start_time": "16:45", "end_time": "18:00", "talks": [ { "title": "Closed-Loop autonomous discovery of functional membranes and 2D Materials for resource recovery and energy applications", "speaker": "Daria Andreeva", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "talk_id": "invited:Daria Andreeva", "abstract": "The development of composite materials for sustainable separations, resource recovery, and energy technologies is often constrained by the vast multidimensional design spaces associated with composition, structure, processing conditions, and operating environments. Traditional trial-and-error experimentation is increasingly inadequate for navigating these complex parameter spaces and identifying optimal material architectures. Closed-loop autonomous laboratories offer a transformative paradigm by integrating artificial intelligence, automated experimentation, high-throughput characterization, and adaptive decision-making into self-driving discovery platforms.\nIn this talk, I present a vision for the closed-loop autonomous discovery of functional membranes and two-dimensional (2D) materials for resource recovery and energy applications. The proposed framework combines machine learning-driven experiment planning, automated synthesis and fabrication, real-time characterization, and active learning to iteratively optimize material performance with minimal human intervention. Particular emphasis is placed on membrane-based ion separations, critical mineral recovery, electrocatalytic materials, and self-assembled 2D architectures.\nSeveral representative case studies will be discussed, including AI-guided optimization of 2D materials-based membranes for selective lithium extraction from complex brines and battery recycling streams, for $\\ce{CO_2}$ capture and conversion, and closed-loop development of functional materials for selective recovery of precious metals from electronic waste. We further outline opportunities for integrating computer vision, robotic experimentation, digital twins, and autonomous scientific agents to control and optimize non-equilibrium self-assembly processes in real time.\nBy coupling intelligent decision-making with automated experimentation, autonomous laboratories can accelerate the discovery of high-performance materials while reducing experimental co", "speaker_profile_url": "https://cde.nus.edu.sg/mse/staff/andreeva-baeumler-daria/", "section": [ "AI for Materials Science", "AI for Chemistry", "Self-Driving Labs" ] }, { "talk_id": "oral:closed-loop-optimization-of-mono-functionalization-via-suzuki-miyaura-reaction", "title": "Closed-loop Optimization of Mono-functionalization via Suzuki-Miyaura Reaction", "speaker": "Yuuya Nagata", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=gx1gU7QrVT", "openreview_id": "gx1gU7QrVT", "speaker_profile_url": "https://openreview.net/profile?id=~Yuuya_Nagata1" }, { "talk_id": "oral:flow-chemistry-as-a-platform-for-experimental-multi-objective-optimization-of-heterogeneous-polymer-synthesis", "title": "Flow Chemistry as a Platform for Experimental Multi-objective Optimization of Heterogeneous Polymer Synthesis", "speaker": "Nicholas Warren", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=dWUbgH5nwk", "openreview_id": "dWUbgH5nwk", "speaker_profile_url": "https://openreview.net/profile?id=~Nicholas_J_Warren1" }, { "talk_id": "oral:nimo-universal-middleware-for-closed-loop-materials-exploration", "title": "NIMO: Universal Middleware for Closed-Loop Materials Exploration", "speaker": "Ryo Tamura", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=bSKw0tnSx6", "openreview_id": "bSKw0tnSx6", "speaker_profile_url": "https://openreview.net/profile?id=~Ryo_Tamura1" }, { "talk_id": "oral:mcp-enabled-llm-agents-for-closed-loop-optimization-in-real-time-physical-experiments", "title": "MCP-Enabled LLM Agents for Closed-Loop Optimization in Real-Time Physical Experiments", "speaker": "TBC zekun ren", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=WX69yWSjlj", "openreview_id": "WX69yWSjlj", "speaker_display_name": "Jiaen Yee; Hong Zhao Tan; DANNY Zekun Ren" } ] }, { "session_title": "Symmetry-Aware AI for Materials and Physics", "rationale": "Quantum Interactions in Materials: a New Frontier for AI, Generative modeling and tensor-network, Symmetry-Aware Deep Learning for Generalizable STEM Phase Classification, Mapping diverse structures of liquid water and ice using variational autoencoders: A vector quantization approach to discover structural motifs in model latent spaces, VLM4Physics: Equation Discovery Using Multi-modal Inputs", "section": "AI for Physics", "breakout_venue_name": "Morrison", "chair_name": "Shyue Ping Ong", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Marco Bernardi", "title": "Quantum Interactions in Materials: a New Frontier for AI", "speaker": "Marco Bernardi", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "This talk will discuss recent advances at the interface of AI and physics, focusing on tensor learning methods for compressing quantum interactions in materials. We will show that such compression can accelerate first-principles calculations of electronic interactions by orders of magnitude and enable previously inaccessible many-body calculations, including accurate summation of all Feynman diagrams providing a numerically exact solution of the polaron problem. These advances open new directions for predictive calculations of transport phenomena and nonequilibrium dynamics in materials.", "speaker_profile_url": "http://bernardi.caltech.edu/", "section": [ "AI for Physics", "AI for Materials Science" ] }, { "talk_id": "invited:Yuehaw Khoo", "title": "Generative modeling and tensor-network", "speaker": "Yuehaw Khoo", "talk_type": "featured", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract": "Tensor networks have long been a central computational tool in many-body physics, and more recently have emerged as a powerful framework for simulating quantum computers. In this talk, I will describe how tensor networks can also be used as function representations for fast generative modeling, where training and inference can be done through linear algebraic operations. I will also discuss how this opens new opportunities for enhancing physics simulations.", "speaker_profile_url": "https://www.stat.uchicago.edu/~ykhoo/", "section": [ "AI for Physics", "AI for Biology" ] }, { "talk_id": "oral:symmetry-aware-deep-learning-for-generalizable-stem-phase-classification", "title": "Symmetry-Aware Deep Learning for Generalizable STEM Phase Classification", "speaker": "Jiadong Dan", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=lvYSgHsmVx", "openreview_id": "lvYSgHsmVx", "speaker_profile_url": "https://openreview.net/profile?id=~Jiadong_Dan1" }, { "talk_id": "oral:mapping-diverse-structures-of-liquid-water-and-ice-using-variational-autoencoders-a-vector-quantization-approach-to-discover-structural-motifs-in-model-latent-spaces", "title": "Mapping diverse structures of liquid water and ice using variational autoencoders: A vector quantization approach to discover structural motifs in model latent spaces", "speaker": "Yue Yifei", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=BIzubF6L4J", "openreview_id": "BIzubF6L4J", "speaker_profile_url": "https://openreview.net/profile?id=~Yifei_Yue2" }, { "talk_id": "oral:vlm4physics-equation-discovery-using-multi-modal-inputs", "title": "VLM4Physics: Equation Discovery Using Multi-modal Inputs", "speaker": "Qianshu Ye", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=4ZsQWCcyL2", "openreview_id": "4ZsQWCcyL2", "speaker_profile_url": "https://openreview.net/profile?id=~Ye_Qianshu1" } ], "chair_profile_url": "https://materialyze.ai" }, { "session_title": "AI-Driven Unconventional Computing Hardware", "rationale": "CMOS-Integrated Silicon-Oxide Memristors: Reliability Characterization, SPICE-Based Circuit Simulation and potential application in neuromorphic computing, LLM-Powered Autonomous Agents for Spintronic Device Optimization: From Rule-Based to AI-Driven Design, Design Methodologies for Skyrmion-Based Circuits and Systems in AI-Driven Applications: Bi-Directional Integration, Ultra-low-energy skyrmion-based learning automata element for adaptive edge intelligence, Spectrum-Aware Quantum Control beyond Classical Spectral Access", "section": "Unconventional Computing", "breakout_venue_name": "Hullet", "chair_name": "Giacomo Indiveri", "chair_profile_url": "https://www.ini.uzh.ch/", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Fernando Aguirre", "title": "CMOS-Integrated Silicon-Oxide Memristors: Reliability Characterization, SPICE-Based Circuit Simulation and potential application in neuromorphic computing", "speaker": "Fernando Aguirre", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Memristive devices are promising candidates for future embedded non-volatile memory technologies, particularly as conventional embedded Flash faces increasing scaling and integration challenges in advanced CMOS nodes. Silicon-oxide memristors are especially attractive in this context because their material system is closely aligned with silicon CMOS technology, avoiding the need to introduce additional non-standard switching materials into the integration flow. This makes them a relevant platform for exploring dense, scalable and manufacturable memory solutions, while also offering prospects for emerging computing paradigms such as neuromorphic hardware. This talk presents CMOS-integrated silicon-oxide memristors implemented in a arrays of 1T1R cells, where each resistive switching element is addressed by a CMOS selector transistor. Their electrical behaviour is assessed under realistic operating conditions, with emphasis on retention and endurance as key reliability metrics for embedded memory operation. Retention measurements evaluate the stability of programmed resistance states over time, while endurance measurements assess repeated switching between high- and low-resistance states, providing insight into the practical operating window of the devices. These results are discussed in terms of their relevance for reliable memory programming, readout and cycling in circuit environments. In parallel, the talk introduces a SPICE-compatible compact model based on the memdiode framework, where current transport and memory-state evolution are described through coupled equations fitted to experimental data. The model reproduces DC and pulsed operation and enables circuit-level simulations of write, read and cycling schemes in hybrid CMOS-memristor circuits. This provides a link between measured device behaviour and circuit design, allowing the impact of device properties, variability and programming conditions to be explored before implementation. Finally, the talk will b", "speaker_profile_url": "https://scholar.google.es/citations?user=rdhFxZ4AAAAJ&hl=es", "section": [ "Unconventional Computing" ] }, { "talk_id": "oral:llm-powered-autonomous-agents-for-spintronic-device-optimization-from-rule-based-to-ai-driven-design", "title": "LLM-Powered Autonomous Agents for Spintronic Device Optimization: From Rule-Based to AI-Driven Design", "speaker": "Santhosh Sivasubramani", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=pHXce1SbJa", "openreview_id": "pHXce1SbJa", "speaker_profile_url": "https://openreview.net/profile?id=~Santhosh_Sivasubramani1" }, { "talk_id": "oral:design-methodologies-for-skyrmion-based-circuits-and-systems-in-ai-driven-applications-bi-directional-integration", "title": "Design Methodologies for Skyrmion-Based Circuits and Systems in AI-Driven Applications: Bi-Directional Integration", "speaker": "Santhosh Sivasubramani", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=EW3l9iSotL", "openreview_id": "EW3l9iSotL", "speaker_profile_url": "https://openreview.net/profile?id=~Santhosh_Sivasubramani1" }, { "talk_id": "oral:ultra-low-energy-skyrmion-based-learning-automata-element-for-adaptive-edge-intelligence", "title": "Ultra-low-energy skyrmion-based learning automata element for adaptive edge intelligence", "speaker": "Santhosh Sivasubramani", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=0xEhro1zlW", "openreview_id": "0xEhro1zlW", "speaker_profile_url": "https://openreview.net/profile?id=~Santhosh_Sivasubramani1" }, { "talk_id": "oral:spectrum-aware-quantum-control-beyond-classical-spectral-access", "title": "Spectrum-Aware Quantum Control beyond Classical Spectral Access", "speaker": "Jianlong Lu", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=dC3ln7TA2h", "openreview_id": "dC3ln7TA2h", "speaker_profile_url": "https://openreview.net/profile?id=~Jianlong_Lu1" } ] }, { "session_title": "Responsible AI, Cultural Alignment, and Social Trust", "rationale": "Beyond Alignment: Grounding AI in Society, Exploring Social Trust in AI, Align AI to our Aspirations, not our Flaws, Ethical Nail Salons: A community-governed and SDL-facilitated approach to mitigate occupational chemical hazards in nail salons, AI & Culture Alignment: Interpretation over Measurement", "section": "AI for Society", "breakout_venue_name": "Moor", "chair_name": "Truyen Tran", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Nancy F. Chen", "title": "Beyond Alignment: Grounding AI in Society", "speaker": "Nancy F. Chen", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Modern AI systems are increasingly aligned to human instructions and preferences. Yet alignment alone does not guarantee that AI understands the people, cultures, contexts, and realities in which it operates. A model can be helpful, harmless, and fluent—and still fail to connect meaningfully with society.\nThis challenge emerges from a broader evolution of AI—from symbolic representations and curated annotations, to statistical learning, large-scale representation learning, and alignment—each expanding what machines can learn from human-generated data.\nThis talk argues that the next frontier of AI is grounding: connecting AI not only to language and data, but also to human perception, interaction, emotions, culture, and domain knowledge. Grounding shifts the focus from generating plausible outputs to understanding context, adapting to diverse communities, and supporting real-world decisions.\nDrawing on examples from speech, language, and multimodal AI, I will illustrate how grounding can be achieved through human interaction, multicultural reasoning, and real-world deployment. Case studies include MERaLiON, the first multimodal large language model for Southeast Asia; SingaKids AI Tutor, a multilingual learning companion for children learning Malay, Mandarin, and Tamil; and Siu Dai, a telebot supporting patients in chronic disease management.\nAs AI becomes embedded in education, healthcare, government, and everyday life, the challenge is no longer merely to align AI with human preferences, but to ground AI in human experience. The future of AI in society depends on it.", "speaker_profile_url": "https://www.a-star.edu.sg/cfar/about-cfar/our-team/dr-nancy-f-chen", "section": [ "AI for Society", "Ethical Approaches to AI", "AI for Education and Policy" ] }, { "talk_id": "invited:Sulfikar Amir", "title": "Exploring Social Trust in AI", "speaker": "Sulfikar Amir", "talk_type": "featured", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract": "What drives people to have trust in using artificial intelligence (AI)? How does the institutional environment shape social trust in AI? This talk addresses these questions to explain the role of institutions in allowing AI-based technologies to be socially accepted. In this talk, social trust in AI is situated in three institutional entities, namely, the government, tech companies, and the scientific community. It is posited that the level of social trust in AI is correlated to the level of trust in these institutions. The stronger the trust in the institutions, the deeper the social trust in the use of AI. The talk draws on a cross-country survey in East Asia to show convincing evidence of how institutions shape social trust in AI and its acceptance. It reveals that trust in institutions is positively associated with trust in AI technologies. Trust in institutions is based on perceived competence, benevolence, and integrity. A higher level of trust in AI technologies leads to a higher level of intention to use these technologies. This has profound implications on the governance of AI in society. By taking into account institutional factors in the planning and implementation of AI regulations, we can be assured that social trust in AI is sufficiently founded.", "speaker_profile_url": "https://dr.ntu.edu.sg/entities/person/Sulfikar-Amir", "section": [ "Ethical Approaches to AI" ] }, { "talk_id": "invited:Nikita Kazeev", "title": "Align AI to our Aspirations, not our Flaws", "speaker": "Nikita Kazeev", "talk_type": "featured", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract": "As AI grows more capable and more widely adopted, it quietly becomes one of the largest social-engineering projects in history: billions of people will heed its advice on questions personal and professional. That makes one design choice decisive - which human values, society model, and assumptions we build into it.\n\nThe natural target are human preferences and broader culture. They are accessible, they fit the commercial logic of giving customers what they want, and they wear the color of democratic legitimacy. I will argue this target is a trap. The values evolution and cultural history installed in us were optimized for short-term survival and tribal cohesion, not for the flourishing of large, technologically empowered societies - and they routinely fail us. We reward the yes-man and call sycophancy the joke of our field. We pull strings for our relatives and keep our country corrupt and poor. We chase income, status, and engagement - then lament the solitude. We let moral judgments quietly overwrite empirical ones. Much of the dysfunction in the world is a human-alignment failure; train AI to reproduce our preferences and we amplify it.\n\nBut AI is not only a poison - it can be a cure. The case is not for aligning AI to who we are, but to who we wish we were. Strip away the in-the-moment impulse and what is striking is not that every culture wants identical things, but that every culture distinguishes its impulses from its aspirations - and the aspirations, where they are written down, converge far more than the impulses do: to be honest when a lie is easier (integrity), correct rather than flattering (competence), fair to the stranger as to our own kin (benevolence), faithful to the person we will be in ten years. We know what the better version looks like; we just cannot reliably be it. AI need not carry the baggage that defeats us - no tribe to favor, no ego to defend, no present bias unless we build one in. And for the first time we have the tools to realize these aspirations rather than merely state them: we can inspect an AI's reasoning and re-run it under counterfactuals; we can judge it against outcomes - did the business get built, the patient recover, the friendship survive - instead of against applause; and we can select for virtue, keeping the version that stays honest under pressure, as we rarely manage to do with human power. The better angel of our nature need not be a fiction we tell ourselves. It can be something we specify, measure, and build.", "speaker_profile_url": "https://kazeevn.github.io/", "section": [ "Ethical Approaches to AI", "AI for Society" ] }, { "talk_id": "oral:ethical-nail-salons-a-community-governed-and-sdl-facilitated-approach-to-mitigate-occupational-chemical-hazards-in-nail-salons", "title": "Ethical Nail Salons: A community-governed and SDL-facilitated approach to mitigate occupational chemical hazards in nail salons", "speaker": "Reena Shadaan", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=pAonueE8sA", "openreview_id": "pAonueE8sA", "speaker_profile_url": "https://openreview.net/profile?id=~Reena_Shadaan1" }, { "talk_id": "oral:ai-culture-alignment-interpretation-over-measurement", "title": "AI & Culture Alignment: Interpretation over Measurement", "speaker": "Eric J. W. Orlowski", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=mvSfYxXgMx", "openreview_id": "mvSfYxXgMx", "speaker_profile_url": "https://openreview.net/profile?id=~Eric_J._W._Orlowski1" } ], "chair_profile_url": "https://truyentran.github.io/" } ] } ], "breaks": [ { "name": "Tea Break", "start_time": "10:30", "end_time": "11:00" }, { "name": "Tea Break", "start_time": "16:15", "end_time": "16:45" } ], "poster_session": { "name": "Poster Session #1 / Lunch", "start_time": "12:30", "end_time": "14:45", "session_index": 1 } }, { "date": "2026-06-18", "plenary_talks": [ { "talk_id": "invited:Vivek Natarajan", "title": "General-purpose AI systems from Google DeepMind designed to accelerate scientific discovery and democratize medical expertise", "speaker": "Vivek Natarajan", "talk_type": "plenary", "length": 1, "start_time": "09:00", "end_time": "09:30", "abstract": "First, the AI co-scientist, a multi-agent Gemini-based system, assists researchers by systematically generating and refining novel hypotheses for complex scientific challenges. This approach has yielded promising, lab-validated results, including identifying drugs for repurposing against acute myeloid leukemia, discovering new therapeutic targets for liver fibrosis (Advanced Science), and recapitulating a novel gene transfer mechanism for bacterial resistance (Cell). While early, the co-scientist represents a promising step toward a true collaborative AI partner for scientist.\nSecondly, the AI co-physician, AMIE, aims to give doctors superpowers and make medical expertise universally accessible. In simulated settings, AMIE outperformed primary care physicians across multiple clinical evaluation axes (Nature) and is showing promise as an assistive tool in ongoing real-world validations (Nature).\nTogether, these initiatives demonstrate AI's potential to transform scientific research and care delivery.", "speaker_profile_url": "https://natviv.me", "section": [ "AI Agents and LLMs for Science", "AI for Medicine and Healthcare" ] }, { "talk_id": "invited:Bartosz Grzybowski", "title": "Can robots help us redefine chemical reactions?", "speaker": "Bartosz Grzybowski", "talk_type": "plenary", "length": 1, "start_time": "11:00", "end_time": "11:30", "abstract": "Organic chemistry is undergoing two profound transformations – driven by algorithms and by robots – that are reshaping how we design molecules and discover reactions. Traditionally, tasks such as synthesis planning and reaction discovery have relied on human imagination, intuition and experimental dexterity. Today, however, algorithms can plan syntheses of complex natural products at a level rivaling top human experts, and when paired with robotic platforms, they can systematically uncover new and useful reaction types.\n \nBeyond automation, these technologies challenge our very understanding of chemical reactivity. Rather than viewing reactions as static, single-line equations (e.g., A + B → C), we can now explore them as dynamic networks embedded in multidimensional hyperspaces of conditions. Within these landscapes, control parameters act as switches, enabling transitions between distinct major products. In some cases, these networks rival the complexity of biochemical systems, revealing unexpected product distributions and novel mechanistic pathways.\nThis talk will explore how these algorithmic and robotic revolutions are not only accelerating discovery but also redefining the conceptual foundations of synthetic chemistry.", "speaker_profile_url": "https://grzybowski-group.net", "section": [ "AI for Chemistry", "AI for Materials Science", "Self-Driving Labs" ] }, { "talk_id": "invited:Karsten Reuter", "title": "When the Algorithms Take Over: AI for Experiment Planning and Control", "speaker": "Karsten Reuter", "talk_type": "plenary", "length": 1, "start_time": "12:00", "end_time": "12:30", "abstract": "More performant and durable materials are urgently needed to further drive the transition to a sustainable energy system. Unfortunately, accelerated materials discovery is in this field presently still more claim than practical reality. Computational screening approaches hinge on efficient descriptors that only reflect nominal materials properties of the crystalline bulk, simple bulk-truncated surfaces or idealized lattice-matching interfaces. They can thus not account for the substantial, complex and continuous structural, compositional and morphological transitions at the working surfaces or interfaces of functional materials in catalysts, electrolyzers or batteries. Accelerated experimental discovery in turn still suffers from severe throughput limitations, as easily automatable human steps are rarely limiting the overall workflows. \n\nIn my talk I will illustrate how the convergence of AI-based experiment planning and control with automation and robotization is currently starting to change this picture. Examples will not only cover straightforward loop-style self-driving labs for direct exploration of design spaces, but also approaches geared toward mechanistic understanding like automated reaction mechanism generation or autonomous electron microscopy. Methodological frontiers concern significant or varying noise levels (e.g. in case of multi-fidelity measurements), the design of larger numbers of data points (to meet batch-type operation in increasingly parallelized workflows), or agility to either autonomously adapt the shape and dimensions of the search spaces across loops or react to corresponding changes imposed by human scientists.", "speaker_profile_url": "https://www.fhi.mpg.de/th-department", "section": [ "AI for Chemistry", "AI for Materials Science", "AI for Physics" ] } ], "keynote_talks": [ { "talk_id": "invited:Ray Meng Gao", "title": "New Frontiers in Machine Learned Quantum Chemistry", "speaker": "Ray Meng Gao", "talk_type": "featured", "length": 1, "start_time": "09:30", "end_time": "09:45", "abstract": "The release of foundational models and massive datasets, such as UMA and Omol25, has propelled machine-learned quantum chemistry into uncharted domains. Here we explore new challenges and frontiers in the next phase of atomistic AI models, specifically focusing on bridging the timescale gap between machine learning interatomic potentials (MLIPs) and classical force fields. We also examine areas where scale can continue to unlock new science such as large-scale electronic structure predictions.", "speaker_profile_url": "https://sites.google.com/view/raymgao/about", "section": [ "AI for Chemistry", "AI for Materials Science", "AI for Science" ] }, { "talk_id": "invited:Michele Ceriotti", "title": "Let them learn: AI models that master materials physics", "speaker": "Michele Ceriotti", "talk_type": "featured", "length": 1, "start_time": "09:45", "end_time": "10:00", "abstract": "Machine learning is transforming the way we perform simulations to predict and understand the properties of materials. Traditional \"physics-informed\" models build symmetry, smoothness, and other physical priors into the mathematical structure of the model to guide learning. In this talk, I'll explore how much physics we really need to include when teaching machines the quantum behavior of matter. I'll contrast models constrained by physical assumptions with emerging, unconstrained approaches that learn physical relationships directly from data. These models can reach, and sometimes exceed, the accuracy and efficiency of their physics-based counterparts, though they require some care to avoid unphysical results.\nI'll illustrate these ideas using PET-MAD, a lightweight, data-driven model that learns across the periodic table, providing accurate predictions of the microscopic properties of materials that include a quantification of the model uncertainties, both against the reference electronic-structure calculations, and against experiments.", "speaker_profile_url": "https://people.epfl.ch/michele.ceriotti?lang=en", "section": [ "AI for Materials Science" ] }, { "talk_id": "invited:Ryutaro Uchiyama", "title": "AI-Driven Scaffolding of Open-ended Movement Exploration", "speaker": "Ryutaro Uchiyama", "talk_type": "featured", "length": 1, "start_time": "10:00", "end_time": "10:15", "abstract": "To limit the complexity of motor control, the vertebrate nervous system seeks strategic, low-dimensional motor coordination structures. But when environments are dense with hidden transition dynamics, redundant (over-actuated) control dimensions can serve as the raw material for skill innovation, constituting a fundamental tradeoff for motor learning. I will discuss how humans may have evolved a distinctive strategy to navigate this tradeoff – as suggested by the apparent species-uniqueness of our open-ended behavioral variability. I will then describe my research group's efforts to engineer a gamified platform that uses AI-driven, real-time kinematic analysis to nudge individuals toward higher-dimensional subspaces of movement exploration, thus closing the loop between theory and application.", "speaker_profile_url": "https://ryu.sg", "section": [ "AI for Biology", "AI for Medicine and Healthcare", "AI for Society" ] }, { "talk_id": "invited:Truyen Tran", "title": "The New Scientific Method: Taste, Truth, and Thinking with AI", "speaker": "Truyen Tran", "talk_type": "featured", "length": 1, "start_time": "10:15", "end_time": "10:30", "abstract": "Artificial intelligence is rapidly entering science, not only as a tool for data analysis and automation, but increasingly as a partner in hypothesis generation, literature synthesis, experiment design, and reasoning. Yet the real question remains: Can AI help us do better science? In this talk, I argue that the age of advanced AI requires a rethink of the scientific method itself. Drawing on great thinkers of our time and my own experience, I propose that AI-powered science will change three things: taste (choosing important problems), truth (not fooling ourselves), and thinking (building the right representations and explanations). The central claim is that AI should not merely automate scientific output. Its deeper value lies in augmenting the key cognitive functions of science: identifying important questions, challenging weak explanations, preserving scientific memory, and helping us think more effectively. The future scientist is not human alone or AI alone, but a well-designed human–AI cognitive partnership.", "speaker_profile_url": "https://truyentran.github.io/", "section": [ "AI Agents and LLMs for Science", "AI for Science" ] }, { "talk_id": "invited:Adam Gormley", "title": "Polymer Biomaterials in a Self-Driving Lab", "speaker": "Adam Gormley", "talk_type": "featured", "length": 1, "start_time": "11:30", "end_time": "11:45", "abstract": "The seamless integration of synthetic materials with biological systems remains a grand challenge, often curtailed by the sheer complexity of the cell-material interface. For decades, biomaterial scientists and engineers have designed around this complexity by rationally designing new materials one experiment at a time. However, recent advances in laboratory automation, high throughput analytics, and artificial intelligence / machine learning (AI/ML) now provide a unique opportunity to fully automate the design process. In this talk, we put forth our efforts to develop a self-driving biomaterials lab that can rapidly iterate through design spaces and identify unique material properties that perfectly synergize with biological complexity.", "speaker_profile_url": "https://bme.rutgers.edu/adam-j-gormley", "section": [ "Self-Driving Labs", "AI for Medicine and Healthcare", "AI for Biology" ] }, { "talk_id": "invited:Shyue Ping Ong", "title": "Physics, Scaling and Data in Foundation Potentials", "speaker": "Shyue Ping Ong", "talk_type": "featured", "length": 1, "start_time": "11:45", "end_time": "12:00", "abstract": "Foundation potentials approximate the potential energy surface across the periodic table, enabling atomistic simulations at length, time, and chemical scales far beyond the reach of ab initio methods. This capability positions them as powerful engines for materials discovery and design. In this talk, I will outline my perspective on what constitutes a meaningful path forward for the next generation of foundation potentials. I will argue that sustained progress rests on three pillars: the systematic incorporation of relevant physics, the creation of high-quality and application-driven training datasets, and algorithmic designs that scale efficiently in both time and memory. Together, these elements define the next phase of reliable, fast, and broadly applicable foundation potentials for materials science.", "speaker_profile_url": "https://materialyze.ai", "section": [ "AI for Materials Science" ] } ], "early_morning_chair_name": "Melodie Christensen", "late_morning_chair_name": "Beatrice Soh", "late_morning_chair_profile_url": "https://bwysoh.wixsite.com/sohlab", "time_blocks": [ { "block_name": "afternoon_session", "start_time": "14:45", "end_time": "16:15", "sessions": [ { "session_title": "Machine Learning for Materials Discovery and Properties Prediction", "rationale": "A Disorder-Aware Multi-fidelity Framework for Robust Prediction of Superconducting Critical Temperature, Graph learning metallic glass discovery from Wikipedia, Machine learning reveals transferable rules for predicting grain boundary segregation, Atomic-Level Interpretable Multimodal Graph Neural Network for Predicting Carbon Capture in Metal-Organic Frameworks, Discovery of Sustainable Refrigerants through Physics-Informed RL Fine-Tuning of Sequence Models, CSX Framework for Synthesis-Oriented Generative Materials Discovery", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Artem Maevskiy", "chair_profile_url": "https://openreview.net/profile?id=~Artem_Maevskiy1", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "oral:a-disorder-aware-multi-fidelity-framework-for-robust-prediction-of-superconducting-critical-temperature", "title": "A Disorder-Aware Multi-fidelity Framework for Robust Prediction of Superconducting Critical Temperature", "speaker": "Linh La", "talk_type": "contributed", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract_url": "https://openreview.net/forum?id=d1kzyXxdsL", "openreview_id": "d1kzyXxdsL", "speaker_profile_url": "https://openreview.net/profile?id=~Linh_La1" }, { "talk_id": "oral:graph-learning-metallic-glass-discovery-from-wikipedia", "title": "Graph learning metallic glass discovery from Wikipedia", "speaker": "Yuanchao Hu", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=N7rEY7r91E", "openreview_id": "N7rEY7r91E", "speaker_profile_url": "https://openreview.net/profile?id=~Yuan-Chao_Hu1" }, { "talk_id": "oral:machine-learning-reveals-transferable-rules-for-predicting-grain-boundary-segregation", "title": "Machine learning reveals transferable rules for predicting grain boundary segregation", "speaker": "Jingbei Bai", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=Hbjpt4w3NI", "openreview_id": "Hbjpt4w3NI", "speaker_profile_url": "https://openreview.net/profile?id=~Jingbei_Bai1" }, { "talk_id": "oral:atomic-level-interpretable-multimodal-graph-neural-network-for-predicting-carbon-capture-in-metal-organic-frameworks", "title": "Atomic-Level Interpretable Multimodal Graph Neural Network for Predicting Carbon Capture in Metal-Organic Frameworks", "speaker": "Guangcun Shan", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=3umSoJmSzr", "openreview_id": "3umSoJmSzr", "speaker_profile_url": "https://openreview.net/profile?id=~Guangcun_Shan1" }, { "talk_id": "oral:discovery-of-sustainable-refrigerants-through-physics-informed-rl-fine-tuning-of-sequence-models", "title": "Discovery of Sustainable Refrigerants through Physics-Informed RL Fine-Tuning of Sequence Models", "speaker": "Adrien Goldszal", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=0kyF6VHtzp", "openreview_id": "0kyF6VHtzp", "speaker_profile_url": "https://openreview.net/profile?id=~Adrien_Goldszal1" }, { "talk_id": "oral:csx-framework-for-synthesis-oriented-generative-materials-discovery", "title": "CSX Framework for Synthesis-Oriented Generative Materials Discovery", "speaker": "Shuya Yamazaki", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=R0kAfLUThA", "openreview_id": "R0kAfLUThA", "speaker_profile_url": "https://openreview.net/profile?id=~Shuya_Yamazaki1" } ] }, { "session_title": "Autonomous labs for chemistry and materials discovery", "rationale": "Toward Safe Autonomy in Self-Driving Laboratories, Add, Mix, Heat, Filter - Repeat, Repeat, Repeat, A Self-Driving Lab for Novel Energy Material Discovery, Combining robotic deposition tools and advanced characterization to enable ML-guided material discovery, Printing kinetic data and microkinetic models in an automated lab, An End-to-end, Autonomous Platform for Liquid-liquid Extraction Optimization", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Han Hao", "chair_profile_url": "https://openreview.net/profile?id=~Han_Hao1", "start_time": "14:45", "end_time": "16:15", "talks": [ { "title": "Toward Safe Autonomy in Self-Driving Laboratories", "speaker": "Leong Shi Xuan", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "talk_id": "invited:Leong Shi Xuan", "abstract": "As self-driving laboratories expand the scope and complexity of experimental processes they can perform, ensuring safe operation becomes critical. I will highlight practical limitations of current AI and robotics systems in real laboratory settings and discuss opportunities for building more trustworthy and robust autonomous scientific platforms.", "speaker_profile_url": "https://www.leongshixuan.com/", "section": [ "Self-Driving Labs", "AI for Materials Science", "AI for Chemistry" ] }, { "talk_id": "oral:add-mix-heat-filter-repeat-repeat-repeat", "title": "Add, Mix, Heat, Filter - Repeat, Repeat, Repeat", "speaker": "Christopher Hassam", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=VviEt6sAU4", "openreview_id": "VviEt6sAU4", "speaker_profile_url": "https://openreview.net/profile?id=~Christopher_L._Hassam1" }, { "talk_id": "oral:a-self-driving-lab-for-novel-energy-material-discovery", "title": "A Self-Driving Lab for Novel Energy Material Discovery", "speaker": "Hugo Kvanta", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=K8szkAkoLa", "openreview_id": "K8szkAkoLa", "speaker_profile_url": "https://openreview.net/profile?id=~Hugo_Kvanta1" }, { "talk_id": "oral:combining-robotic-deposition-tools-and-advanced-characterization-to-enable-ml-guided-material-discovery", "title": "Combining robotic deposition tools and advanced characterization to enable ML-guided material discovery", "speaker": "Tim Kodalle", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=EVbI2EVsUK", "openreview_id": "EVbI2EVsUK", "speaker_profile_url": "https://openreview.net/profile?id=~Tim_Kodalle1" }, { "talk_id": "oral:printing-kinetic-data-and-microkinetic-models-in-an-automated-lab", "title": "Printing kinetic data and microkinetic models in an automated lab", "speaker": "Linden Schrecker", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=E491rJ3Pnp", "openreview_id": "E491rJ3Pnp", "speaker_profile_url": "https://openreview.net/profile?id=~Linden_Schrecker1" }, { "talk_id": "oral:an-end-to-end-autonomous-platform-for-liquid-liquid-extraction-optimization", "title": "An End-to-end, Autonomous Platform for Liquid-liquid Extraction Optimization", "speaker": "Maria Politi", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=qmrjNFUJXC", "openreview_id": "qmrjNFUJXC", "speaker_profile_url": "https://openreview.net/profile?id=~Maria_Politi1" } ] }, { "session_title": "Foundation models and LLMs for biology, molecules, and materials", "rationale": "Molecular Foundation Models: from Pretraining-Finetuning to LLMs, Instructing a Chatbot to Design Nucleic Acid Probes for Diagnostics, Local-Global Associative Frames for Symmetry-Preserving Crystal Structure Modeling, Bridging LLM-based planning and workflow languages for automated, validated, scalable exploration of scRNA-seq analyses, From Molecules to Materials and Proteins: Flow Autoencoders as Lossless and Unified Tokenizers, Latent World Models of Cell Painting Data for In Silico Phenotypic Screening", "section": "AI for Biology", "breakout_venue_name": "Morrison", "chair_name": "Yimu Zhao", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Yatao Bian", "title": "Molecular Foundation Models: from Pretraining-Finetuning to LLMs", "speaker": "Yatao Bian", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Molecular foundation models are becoming a key interface between molecular data, domain knowledge, and scientific decision-making. In this talk, I will present our recent line of work on molecular foundation models, progressing from pretraining–finetuning on molecular graphs to language-model-based molecular intelligence. I will first revisit how large-scale self-supervised graph pretraining, represented by GROVER, learns chemically meaningful molecular representations from unlabeled molecules, and how graph topology induced optimal transport improves efficient finetuning under scarce labels. I will then discuss the role of expressive architectures for molecules and biomolecules, including graph transformers, cross-dependent GNNs, hypergraph learning, and symmetry-aware models for protein interactions. Finally, I will describe our recent efforts to connect molecular foundation models with LLMs through hierarchical graph tokenization, and to evaluate LLMs in executable molecular workflows via MolViBench. I will close with ongoing work toward cross-domain tokenization for 3D atomic systems, aiming at a more general foundation model across molecules, proteins, and materials.", "speaker_profile_url": "https://yataobian.com/", "section": [ "AI Agents and LLMs for Science", "AI for Chemistry", "AI for Materials Science" ] }, { "talk_id": "oral:instructing-a-chatbot-to-design-nucleic-acid-probes-for-diagnostics", "title": "Instructing a Chatbot to Design Nucleic Acid Probes for Diagnostics", "speaker": "Hongmin Chen", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=ANLPMapSxS", "openreview_id": "ANLPMapSxS", "speaker_profile_url": "https://openreview.net/profile?id=~Hongmin_Chen2" }, { "talk_id": "oral:local-global-associative-frames-for-symmetry-preserving-crystal-structure-modeling", "title": "Local-Global Associative Frames for Symmetry-Preserving Crystal Structure Modeling", "speaker": "TBC Haowei Hua", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=4V10BQdknT", "openreview_id": "4V10BQdknT", "speaker_display_name": "Wanyu Lin; Haowei Hua" }, { "talk_id": "oral:bridging-llm-based-planning-and-workflow-languages-for-automated-validated-scalable-exploration-of-scrna-seq-analyses", "title": "Bridging LLM-based planning and workflow languages for automated, validated, scalable exploration of scRNA-seq analyses", "speaker": "Yoshinori Hayakawa", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=JTr4zQzTyI", "openreview_id": "JTr4zQzTyI", "speaker_profile_url": "https://openreview.net/profile?id=~Yoshinori_Hayakawa1" }, { "talk_id": "oral:from-molecules-to-materials-and-proteins-flow-autoencoders-as-lossless-and-unified-tokenizers", "title": "From Molecules to Materials and Proteins: Flow Autoencoders as Lossless and Unified Tokenizers", "speaker": "yuxuan Ren", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=8HPvEp2R1e", "openreview_id": "8HPvEp2R1e", "speaker_profile_url": "https://openreview.net/profile?id=~Yuxuan_Ren_ustc1" }, { "talk_id": "oral:latent-world-models-of-cell-painting-data-for-in-silico-phenotypic-screening", "title": "Latent World Models of Cell Painting Data for In Silico Phenotypic Screening", "speaker": "Junhan Wang", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=5VtKp6zr5z", "openreview_id": "5VtKp6zr5z", "speaker_profile_url": "https://openreview.net/profile?id=~Junhan_Wang6" } ], "chair_profile_url": "https://acceleration.utoronto.ca/people/yimu-zhao" }, { "session_title": "AI for Chemical Reaction Prediction and Quantum Chemistry", "rationale": "Learning Arrow Pushing for Reaction Space Prediction and Exploration, Hybrid Computational Strategy for Predicting Complex Ligand–Metal Architectures, Predictive mass spectrometry from quantum-mechanical fragmentation and intensity modelling, ReactionEye: Integrating GC–MS Data and Chemical Context for Multimodal Structure Elucidation in Reaction Screening, Towards Data-Driven Nonlocal Density Functionals: Deep Learning DFT with Attention to approach Chemical Accuracy, Nanostructured Material Design via a Retrieval-Augmented Generation (RAG) Approach: Bridging Laboratory Practice and Scientific Literature", "section": "AI for Chemistry", "breakout_venue_name": "Moor", "chair_name": "Benjamin Chen", "chair_profile_url": "https://research.a-star.edu.sg/researcher/benjamin-chen/", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "oral:learning-arrow-pushing-for-reaction-space-prediction-and-exploration", "title": "Learning Arrow Pushing for Reaction Space Prediction and Exploration", "speaker": "Kye Sung Park", "talk_type": "contributed", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract_url": "https://openreview.net/forum?id=zlj8j2iFg3", "openreview_id": "zlj8j2iFg3", "speaker_profile_url": "https://openreview.net/profile?id=~Kye_Sung_Park1" }, { "talk_id": "oral:hybrid-computational-strategy-for-predicting-complex-ligandmetal-architectures", "title": "Hybrid Computational Strategy for Predicting Complex Ligand–Metal Architectures", "speaker": "Galymzhan Moldagulov", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=vkh5iQNJcH", "openreview_id": "vkh5iQNJcH", "speaker_profile_url": "https://openreview.net/profile?id=~Galymzhan_Moldagulov1" }, { "talk_id": "oral:predictive-mass-spectrometry-from-quantum-mechanical-fragmentation-and-intensity-modelling", "title": "Predictive mass spectrometry from quantum-mechanical fragmentation and intensity modelling", "speaker": "Victor Posligua", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=TG3TqsSCEf", "openreview_id": "TG3TqsSCEf", "speaker_profile_url": "https://openreview.net/profile?id=~Victor_Posligua1" }, { "talk_id": "oral:reactioneye-integrating-gcms-data-and-chemical-context-for-multimodal-structure-elucidation-in-reaction-screening", "title": "ReactionEye: Integrating GC–MS Data and Chemical Context for Multimodal Structure Elucidation in Reaction Screening", "speaker": "Maik Gabriel Niedziella", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=LCWxcL9vnU", "openreview_id": "LCWxcL9vnU", "speaker_profile_url": "https://openreview.net/profile?id=~Maik_Gabriel_Niedziella1" }, { "talk_id": "oral:towards-data-driven-nonlocal-density-functionals-deep-learning-dft-with-attention-to-approach-chemical-accuracy", "title": "Towards Data-Driven Nonlocal Density Functionals: Deep Learning DFT with Attention to approach Chemical Accuracy", "speaker": "Alexander Ryabov", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=KPnVQWFHTb", "openreview_id": "KPnVQWFHTb", "speaker_profile_url": "https://openreview.net/profile?id=~Alexander_Ryabov1" }, { "talk_id": "oral:nanostructured-material-design-via-a-retrieval-augmented-generation-rag-approach-bridging-laboratory-practice-and-scientific-literature", "title": "Nanostructured Material Design via a Retrieval-Augmented Generation (RAG) Approach: Bridging Laboratory Practice and Scientific Literature", "speaker": "Ekaterina Skorb", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=mC818000qd", "openreview_id": "mC818000qd", "speaker_profile_url": "https://openreview.net/profile?id=~Ekaterina_V._Skorb1" } ] }, { "session_title": "AI for Earth System Modeling and Climate Forecasting", "rationale": "NVIDIA Earth-2 and Generative AI for Climate and Weather, Hybrid Physics-AI Digital Twins of the Earth System, From Global to Local: AI-based Climate Downscaling for Southeast Asia, Synthetic Geology: Structural Geology Meets Deep Learning, Physics-informed Deep Operator Networks for Real-Time Spatiotemporal Monitoring of Indoor Air Quality, AI-Enabled 3D Glare Assessment Framework for Urban Solar Planning", "section": "AI for Earth Science", "breakout_venue_name": "Hullet", "chair_name": "Tapio Schneider", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Jeff Adie", "title": "NVIDIA Earth-2 and Generative AI for Climate and Weather", "speaker": "Jeff Adie", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Improved Climate and Weather models are essential for humanity in a changing climate that causes more extreme weather events. Generative AI (GenAI) has become extremely popular in recent times with the advent of LLMs like ChatGPT and image/video generation tools like Stable Diffusion, DALL-E, etc. GenAI methods have also shown great promise in the Climate and Weather domain. Early work with GANs has been augmented with newer generative methods, such as denoising diffusion models and score-based generative models. This talk introduces the NVIDIA Earth-2 program for climate and weather modelling, discusses the features of these latest models and their potential applications. We will also share the latest trends in generative AI for climate and weather modelling.", "speaker_profile_url": "https://developer.nvidia.com/blog/author/jadie/", "section": [ "AI for Earth Science" ] }, { "talk_id": "invited:Gianmarco Mengaldo", "title": "Hybrid Physics-AI Digital Twins of the Earth System", "speaker": "Gianmarco Mengaldo", "talk_type": "featured", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract": "We explore (i) how we can use pre-existing knowledge (e.g., physics) to improve AI systems, and (ii) how we can possible extract some knowledge from AI systems. On the first topic, we present a novel physics-enhanced deep learning hybrid method, namely CondensNet, for resolving cloud physics in general circulation models. On the second topic, we present some results on the use of explainable AI for Earth System applications in the context of extreme events. We conclude with an overall perspective bridging the two topics and grounded in the digital twin paradigm.", "speaker_profile_url": "https://www.mathexlab.com/", "section": [ "AI for Earth Science" ] }, { "talk_id": "oral:from-global-to-local-ai-based-climate-downscaling-for-southeast-asia", "title": "From Global to Local: AI-based Climate Downscaling for Southeast Asia", "speaker": "Juntao Yang", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=LpuEt6VO0S", "openreview_id": "LpuEt6VO0S", "speaker_profile_url": "https://openreview.net/profile?id=~Juntao_Yang3" }, { "talk_id": "oral:synthetic-geology-structural-geology-meets-deep-learning", "title": "Synthetic Geology: Structural Geology Meets Deep Learning", "speaker": "George Turkiyyah", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=WOU334duCN", "openreview_id": "WOU334duCN", "speaker_profile_url": "https://openreview.net/profile?id=~George_Turkiyyah2" }, { "talk_id": "oral:physics-informed-deep-operator-networks-for-real-time-spatiotemporal-monitoring-of-indoor-air-quality", "title": "Physics-informed Deep Operator Networks for Real-Time Spatiotemporal Monitoring of Indoor Air Quality", "speaker": "Nguyen Minh", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=rKeBFcyuZw", "openreview_id": "rKeBFcyuZw", "speaker_profile_url": "https://openreview.net/profile?id=~Nguyen_Tran_Nhat_Minh1" }, { "title": "AI-Enabled 3D Glare Assessment Framework for Urban Solar Planning", "speaker": "Huixuan Sun", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "openreview_id": "ptTeYuE8pi", "talk_id": "oral:ai-enabled-3d-glare-assessment-framework-for-urban-solar-planning", "abstract_url": "https://openreview.net/forum?id=ptTeYuE8pi", "speaker_profile_url": "https://openreview.net/profile?id=~SUN_HUIXUAN1" } ], "chair_profile_url": "https://clima.caltech.edu/" } ] }, { "block_name": "evening_session", "start_time": "16:45", "end_time": "18:00", "sessions": [ { "session_title": "AI-Driven Materials Discovery and Agentic Workflows", "rationale": "Building a Decision-Driven Materials Discovery Institute: Early Insights from MDRI, CatMaster: An Agentic Autonomous System for Computational Heterogeneous Catalysis Research, Exploration and simulation of emergent magnetic materials via AI-driven workflows, Materealize: a multi-agent deliberation system for end-to-end material design and synthesis, SemiMat: A Semi-Supervised Toolkit for Data-Scarce Materials Property Prediction", "section": "AI for Materials Science", "breakout_venue_name": "Sophia", "chair_name": "Santiago Miret", "chair_profile_url": "https://linkedin.com/in/santiago-miret", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "oral:building-a-decision-driven-materials-discovery-institute-early-insights-from-mdri", "title": "Building a Decision-Driven Materials Discovery Institute: Early Insights from MDRI", "speaker": "Stuart Miller", "talk_type": "contributed", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract_url": "https://openreview.net/forum?id=xTBoW5kUWe", "openreview_id": "xTBoW5kUWe", "speaker_profile_url": "https://openreview.net/profile?id=~Stuart_R_Miller1" }, { "talk_id": "oral:catmaster-an-agentic-autonomous-system-for-computational-heterogeneous-catalysis-research", "title": "CatMaster: An Agentic Autonomous System for Computational Heterogeneous Catalysis Research", "speaker": "Honghao Chen", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=o530T2A9oE", "openreview_id": "o530T2A9oE", "speaker_profile_url": "https://openreview.net/profile?id=~Honghao_Chen3" }, { "talk_id": "oral:exploration-and-simulation-of-emergent-magnetic-materials-via-ai-driven-workflows", "title": "Exploration and simulation of emergent magnetic materials via AI-driven workflows", "speaker": "Dorye Luis Esteras", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=gnsadQGcf1", "openreview_id": "gnsadQGcf1", "speaker_profile_url": "https://openreview.net/profile?id=~Dorye_L._Esteras1" }, { "talk_id": "oral:materealize-a-multi-agent-deliberation-system-for-end-to-end-material-design-and-synthesis", "title": "Materealize: a multi-agent deliberation system for end-to-end material design and synthesis", "speaker": "Jaehwan Choi", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=ISXdzKNL8P", "openreview_id": "ISXdzKNL8P", "speaker_profile_url": "https://openreview.net/profile?id=~Jaehwan_Choi1" }, { "talk_id": "oral:semimat-a-semi-supervised-toolkit-for-data-scarce-materials-property-prediction", "title": "SemiMat: A Semi-Supervised Toolkit for Data-Scarce Materials Property Prediction", "speaker": "Yizhe Chen", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=i4EkO5cpxz", "openreview_id": "i4EkO5cpxz", "speaker_profile_url": "https://openreview.net/profile?id=~陈奕哲1" } ] }, { "session_title": "Self-Driving Labs for Materials and Chemical Discovery", "rationale": "Towards in silico prediction of solid-state material synthesizability, Beyond Point Sampling: Autonomous Phase Mapping of Biologic Formulation Stability via Hierarchical and Manifold Active Learning, Self-Driving Labs for Nanomaterials Development for Energy Applications: Syn-thesis, Dispersion, and Composite Forming, An Autonomous Discovery Platform for Inorganic Photovoltaic Absorbers Beyond Lead Halide Perovskites, E-MAP – A Self Driving Lab for Solution Based Combinatorial Semiconductor Discovery", "section": "Self-Driving Labs", "breakout_venue_name": "Moor", "chair_name": "Owen Melville", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Peichen Zhong", "title": "Towards in silico prediction of solid-state material synthesizability", "speaker": "Peichen Zhong", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Dr. Peichen Zhong is an Assistant Professor at the Department of Materials Science and Engineering, National University of Singapore. He obtained a BS in Physics from the University of Science and Technology of China (USTC) in 2018, followed by a PhD in Materials Science from UC Berkeley in 2023. He then completed the postdoctoral work at Lawrence Berkeley National Laboratory (LBNL) and Bakar Institute of Digital Materials for the Planet (BIDMaP). He was awarded the 2023 Rising Stars in Materials Science and Engineering by CMU/MIT/Stanford, the BIDMaP Emerging Scholar Fellowship from the College of Data Science, Computing and Society (CDSS) at UC Berkeley, and the AI2050 Early Career Fellowship by Schmidt Sciences.", "speaker_profile_url": "https://zhongpc.github.io/", "section": [ "AI for Materials Science", "Self-Driving Labs", "AI for Chemistry" ] }, { "talk_id": "oral:beyond-point-sampling-autonomous-phase-mapping-of-biologic-formulation-stability-via-hierarchical-and-manifold-active-learning", "title": "Beyond Point Sampling: Autonomous Phase Mapping of Biologic Formulation Stability via Hierarchical and Manifold Active Learning", "speaker": "Kiran Vaddi", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=lOD9GSWPTz", "openreview_id": "lOD9GSWPTz", "speaker_profile_url": "https://openreview.net/profile?id=~Kiran_Vaddi1" }, { "talk_id": "oral:self-driving-labs-for-nanomaterials-development-for-energy-applications-syn-thesis-dispersion-and-composite-forming", "title": "Self-Driving Labs for Nanomaterials Development for Energy Applications: Syn-thesis, Dispersion, and Composite Forming", "speaker": "Shun Muroga", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=jFWPPuexzN", "openreview_id": "jFWPPuexzN", "speaker_profile_url": "https://openreview.net/profile?id=~Shun_Muroga1" }, { "talk_id": "oral:an-autonomous-discovery-platform-for-inorganic-photovoltaic-absorbers-beyond-lead-halide-perovskites", "title": "An Autonomous Discovery Platform for Inorganic Photovoltaic Absorbers Beyond Lead Halide Perovskites", "speaker": "Udo Bach", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=bt81N6FqVw", "openreview_id": "bt81N6FqVw", "speaker_profile_url": "https://openreview.net/profile?id=~Udo_Bach1" }, { "talk_id": "oral:e-map-a-self-driving-lab-for-solution-based-combinatorial-semiconductor-discovery", "title": "E-MAP – A Self Driving Lab for Solution Based Combinatorial Semiconductor Discovery", "speaker": "Holger RöHM", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=9X0wVYxvlF", "openreview_id": "9X0wVYxvlF", "speaker_profile_url": "https://openreview.net/profile?id=~Holger_Röhm1" } ], "chair_profile_url": "https://openreview.net/profile?id=~Owen_Alfred_Melville1" }, { "session_title": "AI for Chemistry: Molecular Design, Reaction Prediction, and Autonomous Synthesis", "rationale": "CatPlat: A Physics-Grounded Data Engine for AI-Driven Materials Discovery, SynTwins: A Retrosynthesis-Guided Framework for Synthesizable Molecular Analog Generation, Yield Prediction of Organic Reactions in Biased Datasets via Positive-Unlabeled Learning, Generative Design and Experimental Validation of Non-Fullerene Acceptors for Photovoltaics, Adaptive Human-in-the-Loop Optimization Using Language-Guided Priors for Chemical Synthesis", "section": "AI for Chemistry", "breakout_venue_name": "Hullet", "chair_name": "Yanwei Lum", "chair_profile_url": "https://cde.nus.edu.sg/chbe/staff/lum-yanwei/", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Benjamin Chen", "title": "CatPlat: A Physics-Grounded Data Engine for AI-Driven Materials Discovery", "speaker": "Benjamin Chen", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "AI-driven materials discovery platforms can now propose candidate materials faster than traditional evidence pipelines can reliably evaluate them. This creates a new bottleneck: generating trustworthy physical evidence fast enough to guide models, experiments, and subsequent decisions. In this talk, I will present CatPlat, a high-throughput simulations platform to generate high-fidelity data for validating new materials. CatPlat provides reusable computational building blocks for constructing, executing, and analysing large-scale simulation campaigns, while recording detailed provenance for every calculation. This makes it a versatile and verifiable data generation platform for training downstream models and guiding experimental decisions. I will demonstrate the utility of CatPlat by showing how it enabled closed-loop discovery of novel $\\ce{CO_2}$-to-methanol catalysts with a 95% improved yield within six months.", "speaker_profile_url": "https://research.a-star.edu.sg/researcher/benjamin-chen/", "section": [ "AI for Chemistry" ] }, { "talk_id": "oral:syntwins-a-retrosynthesis-guided-framework-for-synthesizable-molecular-analog-generation", "title": "SynTwins: A Retrosynthesis-Guided Framework for Synthesizable Molecular Analog Generation", "speaker": "Shuan Chen", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=kWZcgn1tmX", "openreview_id": "kWZcgn1tmX" }, { "talk_id": "oral:yield-prediction-of-organic-reactions-in-biased-datasets-via-positive-unlabeled-learning", "title": "Yield Prediction of Organic Reactions in Biased Datasets via Positive-Unlabeled Learning", "speaker": "Jan Christopher Spies", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=cmabemXdMd", "openreview_id": "cmabemXdMd", "speaker_profile_url": "https://openreview.net/profile?id=~Jan_Christopher_Spies1" }, { "talk_id": "oral:generative-design-and-experimental-validation-of-non-fullerene-acceptors-for-photovoltaics", "title": "Generative Design and Experimental Validation of Non-Fullerene Acceptors for Photovoltaics", "speaker": "Balamurugan Ramalingam", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=VapxmV5dix", "openreview_id": "VapxmV5dix", "speaker_profile_url": "https://openreview.net/profile?id=~Balamurugan_Ramalingam1" }, { "talk_id": "oral:adaptive-human-in-the-loop-optimization-using-language-guided-priors-for-chemical-synthesis", "title": "Adaptive Human-in-the-Loop Optimization Using Language-Guided Priors for Chemical Synthesis", "speaker": "TBC Amirreza Mottafegh", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=Gni4MCry6X", "openreview_id": "Gni4MCry6X", "speaker_display_name": "Amirreza Mottafegh" } ] }, { "session_title": "Physics-Informed AI and Scientific Computing", "rationale": "Neural network methods for non-smooth PDE-constrained optimization, NeCLO: Neural Convolutional Learning Optimizer for Electromagnetics, AI for Plasma Diagnostics in Laboratory Astrophysics: Reconstructing Invisible Fields from Proton Images, Quantum AI, Transformers with Physics-informed encodings and Simulation-Based inference for robust Gravitational-Wave detection in Pulsar Timing Array data", "section": "ML Algorithmic Advances", "breakout_venue_name": "Olivia", "chair_name": "Eun-Ah Kim", "chair_profile_url": "https://en.wikipedia.org/wiki/Eun-Ah_Kim", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Yongcun Song", "title": "Neural network methods for non-smooth PDE-constrained optimization", "speaker": "Yongcun Song", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "We present neural network methods for solving non-smooth PDE-constrained optimization problems. Our investigation focuses on three challenging categories: (1) optimization with non-smooth regularization, (2) optimal control of PDEs involving interfaces, and (3) optimal control of elliptic variational inequalities. For each category, we develop tailored neural network algorithms that exploit the specific mathematical structure of the underlying problem. The principal advantages of our methods are that they are mesh-free, thus avoiding grid generation challenges; computationally scalable to high dimensions and complex domains; and straightforward to implement. Extensive numerical experiments demonstrate their computational efficiency and accuracy on benchmark problems.", "speaker_profile_url": "https://dr.ntu.edu.sg/entities/person/Yongcun-Song", "section": [ "ML Algorithmic Advances" ] }, { "talk_id": "oral:neclo-neural-convolutional-learning-optimizer-for-electromagnetics", "title": "NeCLO: Neural Convolutional Learning Optimizer for Electromagnetics", "speaker": "Yueming Lyu", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=ZGNtNdLSDk", "openreview_id": "ZGNtNdLSDk", "speaker_profile_url": "https://openreview.net/profile?id=~Yueming_Lyu1" }, { "talk_id": "oral:ai-for-plasma-diagnostics-in-laboratory-astrophysics-reconstructing-invisible-fields-from-proton-images", "title": "AI for Plasma Diagnostics in Laboratory Astrophysics: Reconstructing Invisible Fields from Proton Images", "speaker": "Chun-Sung Jao", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=I3w5pNFhb2", "openreview_id": "I3w5pNFhb2", "speaker_profile_url": "https://openreview.net/profile?id=~Chun-Sung_Jao1" }, { "talk_id": "extra:contributed:quantum-ai:1", "title": "Quantum AI", "speaker": "Nagendra Nagaraja, QpiAI India", "talk_type": "sponsor", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract": "Quantum AI represent the next generation of computing, where quantum technology meets artificial intelligence to unlock the unprecedented computational power. It enables faster analysis smarter predictions and breakthrough innovation across industries, transforming how complex challenges are solved. The talk will focus on how Quantum AI and Autonomous Labs can power smarter labs, accelerate simulations, optimize experiments, and enable autonomous scientific discovery workflows across materials, chemistry, pharma, and deep-tech R&D.", "fixed_slot_start": "17:30", "fixed_slot_end": "17:45", "speaker_profile_url": "https://qpiai.tech/" }, { "talk_id": "oral:transformers-with-physics-informed-encodings-and-simulation-based-inference-for-robust-gravitational-wave-detection-in-pulsar-timing-array-data", "title": "Transformers with Physics-informed encodings and Simulation-Based inference for robust Gravitational-Wave detection in Pulsar Timing Array data", "speaker": "Subhajit Dandapat", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=Bpo4HgB6by", "openreview_id": "Bpo4HgB6by", "speaker_profile_url": "https://openreview.net/profile?id=~Subhajit_Dandapat1" } ] }, { "session_title": "AI for Chemistry and Materials Discovery", "rationale": "Accelerating AI-Powered Chemistry and Materials Science Simulations with NVIDIA ALCHEMI Toolkit, AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement, Toward GPU-Native Electronic Structure Calculations, Ultrafast Spectroscopy Meets Data-Driven Materials Discovery at the Institut Courtois, Université de Montréal, The Open Catalyst 2025 (OC25) Dataset and Models for Solid-Liquid Interfaces", "section": "AI for Materials Science", "breakout_venue_name": "Morrison", "chair_name": "Karsten Reuter", "chair_profile_url": "https://www.fhi.mpg.de/th-department", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Wen Jie Ong", "title": "Accelerating AI-Powered Chemistry and Materials Science Simulations with NVIDIA ALCHEMI Toolkit", "speaker": "Wen Jie Ong", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Machine learning interatomic potentials (MLIPs) are transforming the landscape of computational chemistry and materials science. MLIPs enable atomistic simulations that combine the fidelity of computationally expensive quantum chemistry with the scaling power of AI. Yet, developers and computational chemists working at this intersection face a persistent challenge: a lack of robust, Pythonic tools for GPU-accelerated atomistic simulation.  In this talk, we present NVIDIA ALCHEMI Toolkit and NIM microservices which allow users to flexibly compose atomistic simulation workflows with ease. These simulations workflows are accelerated by optimized batch kernels from ALCHEMI Toolkit-Ops. We will also highlight partner and customer success stories across verticals, from catalysts to beauty products.", "speaker_profile_url": "https://developer.nvidia.com/blog/author/wong", "section": [ "AI for Chemistry", "AI for Materials Science" ] }, { "talk_id": "oral:adsorbflow-energy-conditioned-flow-matching-enables-fast-and-realistic-adsorbate-placement", "title": "AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement", "speaker": "Jiangjie Qiu", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=BE8qgeWmhg", "openreview_id": "BE8qgeWmhg", "speaker_profile_url": "https://openreview.net/profile?id=~JiangJie_Qiu1" }, { "talk_id": "extra:contributed:toward-gpu-native-electronic-structure-calculations:1", "title": "Toward GPU-Native Electronic Structure Calculations", "speaker": "Jiaqi Zheng, SEA Garena", "talk_type": "sponsor", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract": "We present recent progress in the development of a GPU-accelerated Gaussian-type orbital (GTO) integral engine for high-performance electronic structure calculations. Our implementation enables efficient Hartree–Fock and DFT calculations for large molecular systems, achieving speedups of 15–66× over PySCF and 1.5–2.5× over GPU4PySCF in benchmark tests. The integral engine is designed with automatic differentiation in mind, allowing it to support emerging differentiable quantum chemistry workflows. Beyond molecular calculations, the framework is extensible to solid-state systems, including both three-dimensional and two-dimensional materials. These developments provide a flexible and efficient foundation for scalable, differentiable, GPU-native quantum chemistry simulations.", "speaker_profile_url": "https://www.sea.com/products/garena" }, { "talk_id": "extra:contributed:ultrafast-spectroscopy-meets-data-driven-materials-discovery-at-the-institut-courtois-universit-de-montr-al:1", "title": "Ultrafast Spectroscopy Meets Data-Driven Materials Discovery at the Institut Courtois, Université de Montréal", "speaker": "Delphine Bouilly, Institut Courtois, Université de Montreal", "talk_type": "sponsor", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract": "Ultrafast spectroscopy provides direct access to nonequilibrium dynamics in materials, resolving electronic, vibrational, and structural processes on femtosecond timescales. In this talk, I present how these time-resolved measurements can be combined with data-driven approaches to extract meaningful insight from complex datasets, which is a central strategy at the Institut Courtois.\n\nBy integrating machine learning and statistical analysis, we identify patterns, reduce dimensionality, and uncover relationships between structure, dynamics, and function. This interdisciplinary framework connects chemical mechanisms, physical behavior, and computational modeling, enabling more efficient interpretation and guiding materials discovery.\n\nI highlight recent examples demonstrating how the synergy between ultrafast experiments and data-centric methods accelerates understanding and supports the identification of materials with targeted properties, along with key challenges and future directions.\n\nI will also present an overview of opportunities to participate in the scientific program of the Institut Courtois.", "speaker_profile_url": "https://institut-courtois.umontreal.ca/en/team/carlos-silva/" }, { "talk_id": "oral:the-open-catalyst-2025-oc25-dataset-and-models-for-solid-liquid-interfaces", "title": "The Open Catalyst 2025 (OC25) Dataset and Models for Solid-Liquid Interfaces", "speaker": "Nitish Govindarajan", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=4DRmiJJk9w", "openreview_id": "4DRmiJJk9w", "speaker_profile_url": "https://openreview.net/profile?id=~Nitish_Govindarajan1" } ] } ] } ], "breaks": [ { "name": "Tea Break", "start_time": "10:30", "end_time": "11:00" }, { "name": "Tea Break", "start_time": "16:15", "end_time": "16:45" }, { "name": "Conference Dinner", "start_time": "19:00", "end_time": "22:00", "description": "The conference dinner will be held on Thursday, June 18th at the Flower Field Hall in the iconic and award-winning Gardens by the Bay. The evening promises free access to the Flower Dome, light show at the Super Tree Grove, a lovely dinner of international flavours and a cultural show. Celebrate scientific achievements at the poster prize presentation and embrace the opportunity to network and form new scientific relationships. Limited tickets available here.", "location": "Flower Field Hall, Gardens by the Bay, 18 Marina Gardens Dr, Singapore 018953", "location_url": "https://maps.app.goo.gl/7rAXfmEXkM1ngdv97" } ], "poster_session": { "name": "Poster Session #2 / Lunch", "start_time": "12:30", "end_time": "14:45", "session_index": 2 }, "early_morning_chair_profile_url": "https://linkedin.com/in/melodiechristensen" }, { "date": "2026-06-19", "plenary_talks": [ { "talk_id": "invited:Torsten Hoefler", "title": "Can we build an AI Climate Scientist?", "speaker": "Torsten Hoefler", "talk_type": "plenary", "length": 1, "start_time": "09:00", "end_time": "09:30", "abstract": "This talk will focus around how to automate and modernize insights and progress in climate sciences. We will detail our progress on overcoming the critical scalability limits of high-resolution Earth System Modeling (ESM). We will demonstrate how we bridge the gap between \"data-driven\" and \"model-driven\" science in the Age of Computation, specifically targeting the 1km resolution required to resolve convective cloud processes. We present a novel, statistically lossless compression framework (EBCC) that utilizes iterative error-bounding to maintain pixel-wise physical integrity while reducing exabyte-scale storage requirements. Furthermore, we describe our optimization of the ICON model on the GH200 Grace Hopper architecture, where we achieved a 2.4x throughput increase for 1.25km simulations by exploiting CPU-GPU concurrency and data-centric parallel programming. Finally, we conclude with our methodology for full automatic differentiation of legacy Fortran codebases, enabling the seamless integration of neural parameterizations into physics-based dynamical cores through end-to-end backpropagation.", "speaker_profile_url": "https://htor.ethz.ch", "section": [ "AI for Earth Science", "ML Algorithmic Advances" ] }, { "talk_id": "invited:Jacqueline Cole", "title": "Data-Driven Materials Science for Energy-Sustainable Applications", "speaker": "Jacqueline Cole", "talk_type": "plenary", "length": 1, "start_time": "11:00", "end_time": "11:30", "abstract": "This plenary lecture presents the development and application of data-extraction tools and machine-learning (ML) models that are enabling data-driven materials science for energy-sustainable applications. The talk describes how to source high-quality data to build custom databases and apply them to (a) train ML models; (b) realise data-driven materials discovery; (c) equip manufacturing processes with data-driven optimisation strategies; (d) map phase diagrams of materials; (e) forecast trends in materials innovation; (f) create energy-efficient domain-specific language models for materials science to feed agentic AI and help democratise AI for the materials research community. The case studies presented track a theme of energy-sustainable materials.", "speaker_profile_url": "https://en.wikipedia.org/wiki/Jacqui_Cole", "section": [ "AI for Materials Science", "AI for Chemistry" ] }, { "talk_id": "invited:Carlo Vittorio Cannistraci", "title": "Brain-inspired sparse network science for next generation efficient and sustainable AI", "speaker": "Carlo Vittorio Cannistraci", "talk_type": "plenary", "length": 1, "start_time": "12:15", "end_time": "12:45", "abstract": "Artificial neural networks (ANNs) are foundational to contemporary artificial intelligence (AI), however their conventional fully connected architectures are computationally inefficient. Contemporary large language models consume vast amounts of power at rates over 100 times that of the human brain. In stark contrast, the brain's inherently sparse connectivity facilitates exceptional capabilities with minimal expenditure: learning with just a few watts.\n\nBrain-inspired network science research can play a relevant role in designing low-consumption and efficient deep learning. We need to develop concepts and theories for an ecological and sustainable approach to AI. Some of these new computing paradigms can be inspired from the physics of the brain network architecture and its complex systems biology. \n\nAt the Center for Complex Network Intelligence (CCNI) within the Tsinghua Laboratory of Brain and Intelligence (THBI), our research focuses on three pivotal features of brain networks that contribute to efficient computation: (1) Connectivity Sparsity: Implementing sparse connections to reduce computational overhead while maintaining performance; (2) Connectivity Morphology: Exploring the spatial patterns of neural connections to optimize information processing; (3) Neuro-Glia Coupling: Investigating the interactions between neurons and glial cells to enhance computational efficiency.\n\nThis talk will introduce the Cannistraci-Hebb Training soft rule (CHTs), a brain-inspired network science theory that employs a gradient-free approach, relying solely on network topology to predict sparse connectivity during dynamic sparse training. CHTs have demonstrated the potential to achieve ultra-sparse networks with approximately 1% connectivity, outperforming fully connected networks in various tasks.\n\nAdditionally, we will discuss our recent study on the relationship between sparse morphological connectivity and spatiotemporal intelligence. This research introduces neuro", "speaker_profile_url": "https://brain.tsinghua.edu.cn/en/info/1010/1003.htm", "section": [ "Unconventional Computing", "AI for Biology", "ML Algorithmic Advances" ] } ], "keynote_talks": [ { "talk_id": "invited:Alex Hammer", "title": "Beyond the Proof of Concept: Autonomous Electrocatalyst Discovery Within Industrial Constraints", "speaker": "Alex Hammer", "talk_type": "featured", "length": 1, "start_time": "09:30", "end_time": "09:45", "abstract": "Autonomous discovery has shown it can find novel materials in the lab. But a discovery only creates value once it survives the jump from idealized lab conditions to manufacturing reality: device-level performance, robustness, and cost. This lab-to-fab gap is where most efforts stall. We present a six-month autonomous campaign that treats this gap as a design constraint from the outset. Rather than optimizing a simulated proxy, our IRIS platform closes the loop on the full synthesis-to-device sequence, evaluating each candidate as an assembled AEM electrolyzer. Across a deliberately broad 19-element design space, we tested over 800 devices at 30 mV median reproducibility. Where conventional optimization converged, an agentic phase moved past it: an LLM grounded in our experimental database and the patent and literature corpus proposed hypothesis-driven subcampaigns, vetted by human specialists, that extended the Pareto front into new chemical space. The result was three deployable catalysts, the best operating 91 mV below a commercial benchmark. This work exemplifies the broader move from autonomous discovery as academic demonstration toward autonomous research and development as industrial practice.", "section": [ "AI for Materials Science" ] }, { "talk_id": "invited:Felix Hanke", "title": "Orchestrating the Three-Body Problem of Machine Learning, Simulation, and Experiments in Materials Discovery", "speaker": "Felix Hanke", "talk_type": "featured", "length": 1, "start_time": "09:45", "end_time": "10:00", "abstract": "Materials discovery faces a three-body problem: machine learning, physical simulation, and experimentation. Each offers unique strengths but risks missing what the others capture. True acceleration requires orchestrating all three equally. Conway's Law reminds us that system architectures mirror organisational structures; building such workflows requires infrastructure and teams treating simulation, ML, and experimentation as peers. This talk explores the technical and organisational requirements enabling autonomous discovery.\nWe demonstrate how agents coordinate generative design, physics-based validation, and experimental characterisation with scientists maintaining control. Our approach couples GPU-accelerated simulations with adaptive ML models, iterating between in silico predictions and physical measurements. We illustrate this through metal-organic framework design for $\\ce{CO_2}$ capture.", "speaker_profile_url": "https://www.cusp.ai/", "section": [ "AI for Materials Science" ] }, { "talk_id": "invited:Artem Mishchenko", "title": "Why Experimental Data is the Next Frontier for AI x Materials", "speaker": "Artem Mishchenko", "talk_type": "featured", "length": 1, "start_time": "10:00", "end_time": "10:15", "abstract": "While AI has revolutionized our ability to predict materials in silico, the transition from digital discovery to physical reality remains the primary bottleneck. Most existing AI models are trained on computational datasets that often diverge from experimental truth. To unlock the next generation of materials, we must pivot toward the creation of high-fidelity, experimental datasets. This talk discusses the necessity of \"experimental-first\" AI strategies, focusing on how automated laboratories and standardized data capture - including the often-overlooked \"negative results\" - are essential for building models that can truly navigate the complex landscape of physical matter.", "speaker_profile_url": "https://uk.linkedin.com/in/amishche", "section": [ "Self-Driving Labs", "AI for Materials Science" ] }, { "talk_id": "invited:Qianxiao Li", "title": "Learning mesoscopic dynamics", "speaker": "Qianxiao Li", "talk_type": "featured", "length": 1, "start_time": "10:15", "end_time": "10:30", "abstract": "title:: Learning mesoscopic dynamics\n\nabstract:: We discuss some recent work on constructing stable and interpretable mesoscopic dynamics from trajectory data using deep learning. We adopt a hypothesis-driven approach: contrary to using generic neural networks as functional approximators for the vector fields driving a dynamical system or operator based learning approaches, we start with a structured hypothesis class in which well-posedness and numerical stability are a priori ensured. Learning on trajectory data then identifies a specific member of this family suitable for capturing the observed dynamical process, from which scientific predictions and analysis can be made. We demonstrate this approach on a varieity of mesoscopic modelling problems, including the dynamics of spin systems and self-healing in high-entropy alloys.", "speaker_profile_url": "https://blog.nus.edu.sg/qianxiaoli/", "section": [ "AI for Materials Science", "ML Algorithmic Advances", "AI for Physics" ] }, { "talk_id": "invited:Berend Smit", "title": "AI-driven discovery of nanoporous materials", "speaker": "Berend Smit", "talk_type": "featured", "length": 1, "start_time": "11:30", "end_time": "11:45", "abstract": "AI-driven discovery of metal–organic frameworks (MOFs) offers a compelling testbed for accelerating materials innovation, yet it remains fundamentally limited by sparse data and the disconnect between in silico predictions and experimental realization. MOFs are crystalline porous materials built from metal nodes and organic linkers, whose tunable structures enable exceptional performance in applications such as gas separation, carbon capture, and catalysis. \n\nHere, we present a knowledge-integrated AI framework for discovering novel MOFs that combines generative models, physics-based simulations, and application-driven evaluation. Rather than relying solely on data, our approach embeds domain knowledge at every stage: generating chemically valid candidates, predicting adsorption and thermodynamic properties, and evaluating process-level performance using metrics such as net carbon-avoidance cost. This enables efficient exploration of vast MOF design spaces while prioritizing candidates that are not only structurally novel but also functionally relevant. Our results demonstrate that true innovation in MOF discovery arises from integrating AI with chemical insight and process-level objectives, providing a scalable pathway toward autonomous materials discovery systems.", "speaker_profile_url": "https://epfl.ch/labs/lsmo/smit", "section": [ "AI for Materials Science", "AI for Chemistry" ] }, { "talk_id": "invited:Steinn Sigurdsson", "title": "Surviving the Transition: Doing Science in an Age of AI Accelerated Discovery", "speaker": "Steinn Sigurdsson", "talk_type": "featured", "length": 1, "start_time": "11:45", "end_time": "12:00", "abstract": "I discuss the trends in science discovery as traced by publications, and the current impact of generative AI on the process of research and dissemination. We consider some possibilities and pitfalls for the near and medium term.", "speaker_profile_url": "https://www.linkedin.com/in/steinn-sigurdsson-6406398/", "section": [ "AI Agents and LLMs for Science" ] }, { "talk_id": "invited:Tapio Schneider", "title": "First Principles, Fast Algorithms: The Physics-AI Synthesis in Earth System Modeling", "speaker": "Tapio Schneider", "talk_type": "featured", "length": 1, "start_time": "12:45", "end_time": "13:00", "abstract": "While climate change is certain, precisely how climate will change is less clear. But breakthroughs in the accuracy of climate projections and in the quantification of their uncertainties are now within reach, thanks to advances in the computational and data sciences and in the availability of Earth observations from space and from the ground. I will survey the design of a new Earth system model (ESM), developed by the Climate Modeling Alliance (CliMA), which is performance-portable across modern computing architectures. The talk will cover key new concepts and results, including how machine learning (ML) can inform physics-based models with heterogeneous and noisy data and how substantial increases in the accuracy of simulations of uncertain processes can be achieved.", "speaker_profile_url": "https://clima.caltech.edu/", "section": [ "AI for Earth Science" ] }, { "talk_id": "extra:keynote:ai4x-rising-star-awardee-presentation:1", "title": "🏆 AI4X Rising Star Awardee Presentation", "speaker": "AI4X Organizers", "talk_type": "extra_keynote", "length": 1, "start_time": "13:00", "end_time": "13:15", "fixed_slot_start": "13:00", "fixed_slot_end": "13:15", "abstract": "" } ], "early_morning_chair_name": "Simon Billinge", "late_morning_chair_name": "Mohamad Moosavi", "late_morning_chair_profile_url": "https://chem-eng.utoronto.ca/faculty-staff/faculty-members/seyed-mohamad-moosavi/", "midday_chair_name": "Gavin Koon", "time_blocks": [ { "block_name": "afternoon_session", "start_time": "14:45", "end_time": "16:15", "sessions": [ { "session_title": "Machine Learning for Materials Discovery and Design", "rationale": "Physics-Informed AI for Inference and Design under Data Scarcity, Machine Learning-Assisted Search for Skyrmion-Hosting Heterostructures for Device Applications, High-throughput ML Screening of Doped Cathode Active Materials, Discovery of flat-band 2D materials via physics-informed scoring and structure-based learning, Materials informatics framework for accelerated discovery of high-refractive-index 2D materials, Beyond Known Archetypes: A Generative AI Framework for Inverse Design of Flat-Band Materials from Geometric Outliers", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Sonia Azimi Dijvejin", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Seunghwa Ryu", "title": "Physics-Informed AI for Inference and Design under Data Scarcity", "speaker": "Seunghwa Ryu", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Artificial intelligence has shown tremendous potential for scientific and engineering discovery, yet many practical applications remain constrained by sparse measurements, limited simulation data, and the need to satisfy physical laws. In this talk, I will present recent advances in physics-informed AI for inverse estimation and generative design. Topics include uncertainty-aware parameter identification from sparse sensing data using reduced-order physics models, as well as physics-constrained diffusion models for inverse design. These examples demonstrate how embedding physical knowledge into learning and optimization processes can improve data efficiency, reliability, and generalization, providing new opportunities for next-generation scientific and engineering discovery.", "speaker_profile_url": "https://sites.google.com/site/seunghwalab/home?authuser=0", "section": [ "AI for Physics", "ML Algorithmic Advances", "AI for Materials Science" ] }, { "talk_id": "oral:machine-learning-assisted-search-for-skyrmion-hosting-heterostructures-for-device-applications", "title": "Machine Learning-Assisted Search for Skyrmion-Hosting Heterostructures for Device Applications", "speaker": "Sergey Grebenchuk", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=pUSHjW3WXH", "openreview_id": "pUSHjW3WXH", "speaker_profile_url": "https://openreview.net/profile?id=~Sergey_Grebenchuk1" }, { "talk_id": "oral:high-throughput-ml-screening-of-doped-cathode-active-materials", "title": "High-throughput ML Screening of Doped Cathode Active Materials", "speaker": "Mathilde Franckel", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=SP0OKndwpg", "openreview_id": "SP0OKndwpg", "speaker_profile_url": "https://openreview.net/profile?id=~Mathilde_L._D._Franckel1" }, { "talk_id": "oral:discovery-of-flat-band-2d-materials-via-physics-informed-scoring-and-structure-based-learning", "title": "Discovery of flat-band 2D materials via physics-informed scoring and structure-based learning", "speaker": "Xiangwen Wang", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=ReS30YGlC3", "openreview_id": "ReS30YGlC3", "speaker_profile_url": "https://openreview.net/profile?id=~Xiangwen_Wang2" }, { "talk_id": "oral:materials-informatics-framework-for-accelerated-discovery-of-high-refractive-index-2d-materials", "title": "Materials informatics framework for accelerated discovery of high-refractive-index 2D materials", "speaker": "Liudmila Klimova", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=PDkHOSJvos", "openreview_id": "PDkHOSJvos", "speaker_profile_url": "https://openreview.net/profile?id=~Liudmila_Bereznikova1" }, { "talk_id": "oral:beyond-known-archetypes-a-generative-ai-framework-for-inverse-design-of-flat-band-materials-from-geometric-outliers", "title": "Beyond Known Archetypes: A Generative AI Framework for Inverse Design of Flat-Band Materials from Geometric Outliers", "speaker": "Yihao Wei", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=MaiR9lsasm", "openreview_id": "MaiR9lsasm", "speaker_profile_url": "https://openreview.net/profile?id=~Yihao_Wei1" } ] }, { "session_title": "Autonomous Discovery and Society: Geopolitics, Security, and Decentralized Scientific Swarms", "rationale": "The Geopolitics of AI Driven Scientific Discovery: Uneven Geographies of Self Driving Laboratories, Self Driving Discovery of Immersion Cooling Fluids for Data Center, Defending Federated Learning: Adaptive Integration of Differential Privacy, SMPC, and Byzantine Robustness, The Economy of Reasoning: Incentivizing Epistemic Diversity in Decentralized Scientific Swarms, Securing Autonomous Chemical Robots Through Physical and Digital Containment, Edge-AI Driven Automation for Scalable E-Waste Recycling", "section": "AI for Society", "breakout_venue_name": "Hullet", "chair_name": "Kourosh Darvish", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "oral:the-geopolitics-of-ai-driven-scientific-discovery-uneven-geographies-of-self-driving-laboratories", "title": "The Geopolitics of AI Driven Scientific Discovery: Uneven Geographies of Self Driving Laboratories", "speaker": "Julian Prieto", "talk_type": "contributed", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract_url": "https://openreview.net/forum?id=KSIUgeRL7I", "openreview_id": "KSIUgeRL7I", "speaker_profile_url": "https://openreview.net/profile?id=~Julian_Prieto1" }, { "talk_id": "oral:self-driving-discovery-of-immersion-cooling-fluids-for-data-center", "title": "Self Driving Discovery of Immersion Cooling Fluids for Data Center", "speaker": "Maryam Ebrahimiazar", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=Nz5EQhs0ac", "openreview_id": "Nz5EQhs0ac", "speaker_profile_url": "https://openreview.net/profile?id=~Maryam_Ebrahimiazar1" }, { "title": "Defending Federated Learning: Adaptive Integration of Differential Privacy, SMPC, and Byzantine Robustness", "speaker": "Syed Momin Naqvi", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=D0v2L7mjr8", "openreview_id": "D0v2L7mjr8", "talk_id": "oral:defending-federated-learning-adaptive-integration-of-differential-privacy-smpc-and-byzantine-robustness", "speaker_profile_url": "https://openreview.net/profile?id=~Syed_Momin_Naqvi1" }, { "talk_id": "oral:the-economy-of-reasoning-incentivizing-epistemic-diversity-in-decentralized-scientific-swarms", "title": "The Economy of Reasoning: Incentivizing Epistemic Diversity in Decentralized Scientific Swarms", "speaker": "Minh Tri Nguyen", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=CJEsGr3QAx", "openreview_id": "CJEsGr3QAx", "speaker_profile_url": "https://openreview.net/profile?id=~Tri_Minh_Nguyen3" }, { "talk_id": "oral:securing-autonomous-chemical-robots-through-physical-and-digital-containment", "title": "Securing Autonomous Chemical Robots Through Physical and Digital Containment", "speaker": "Dean Thomas", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=9KLyXkwxjA", "openreview_id": "9KLyXkwxjA", "speaker_profile_url": "https://openreview.net/profile?id=~Dean_Thomas1" }, { "talk_id": "oral:edge-ai-driven-automation-for-scalable-e-waste-recycling", "title": "Edge-AI Driven Automation for Scalable E-Waste Recycling", "speaker": "TBC David T T Tran", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=874FRNwQj3", "openreview_id": "874FRNwQj3", "speaker_display_name": "David T T Tran; Ernest E Y Chan" } ], "chair_profile_url": "https://kouroshdarvish.com/" }, { "session_title": "AI for Drug Discovery, Pharma Development, and Clinical Data", "rationale": "Advancing Pharmaceutical Development Through AI/ML-Enabled Experimental Design, MIRA: Medical Time Series Foundation Model for Real-World Health Data, From In Silico Design to Automated Synthesis: An AI-Driven Framework for Late-Stage Functionalization, Agentic AI–Enabled Integration of a Hybrid System of Predictive Models for Accelerated Direct‑Compression Drug Product Development, From Model to Molecule: Rapid Discovery of Potent CDK2 Inhibitors Using Boltz-2, A machine learning workflow to accelerate the design of in vitro release tests from liposomes", "section": "AI for Medicine and Healthcare", "breakout_venue_name": "Morrison", "chair_name": "Nasim Abdollahi", "chair_profile_url": "https://nasimabdollahi.github.io/Research.html", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Melodie Christensen", "title": "Advancing Pharmaceutical Development Through AI/ML-Enabled Experimental Design", "speaker": "Melodie Christensen", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "The proliferation of artificial intelligence and machine learning technologies across the physical sciences has catalyzed a paradigm shift in research methodologies, enabling unprecedented opportunities for research acceleration. Pharmaceutical research organizations are increasingly leveraging these capabilities to transform traditional approaches to drug development.\n\nThis presentation will examine MSD Development Sciences and Clinical Supply's strategic digital transformation initiative, specifically addressing the design, development, and deployment of AI/ML-enabled experimental design platforms within our research ecosystem. We will present our systematic approach to integrating contextualized data, predictive models, optimization algorithms, and advanced experimental planning tools to enhance research efficiency and decision-making accuracy.", "speaker_profile_url": "https://linkedin.com/in/melodiechristensen", "section": [ "Self-Driving Labs", "AI for Medicine and Healthcare", "AI for Biology" ] }, { "talk_id": "oral:mira-medical-time-series-foundation-model-for-real-world-health-data", "title": "MIRA: Medical Time Series Foundation Model for Real-World Health Data", "speaker": "Viktor Schlegel", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=Iih0P8d6f1", "openreview_id": "Iih0P8d6f1", "speaker_profile_url": "https://openreview.net/profile?id=~Viktor_Schlegel1" }, { "talk_id": "oral:from-in-silico-design-to-automated-synthesis-an-ai-driven-framework-for-late-stage-functionalization", "title": "From In Silico Design to Automated Synthesis: An AI-Driven Framework for Late-Stage Functionalization", "speaker": "Lyubomir Kotopanov", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=xBH8I6Oolm", "openreview_id": "xBH8I6Oolm" }, { "talk_id": "oral:agentic-aienabled-integration-of-a-hybrid-system-of-predictive-models-for-accelerated-directcompression-drug-product-development", "title": "Agentic AI–Enabled Integration of a Hybrid System of Predictive Models for Accelerated Direct‑Compression Drug Product Development", "speaker": "Theo Tait", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=tETnWU8L5L", "openreview_id": "tETnWU8L5L" }, { "talk_id": "oral:from-model-to-molecule-rapid-discovery-of-potent-cdk2-inhibitors-using-boltz-2", "title": "From Model to Molecule: Rapid Discovery of Potent CDK2 Inhibitors Using Boltz-2", "speaker": "Sven Papidocha", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=LH0WKMHkIC", "openreview_id": "LH0WKMHkIC", "speaker_profile_url": "https://openreview.net/profile?id=~Sven_M._Papidocha1" }, { "talk_id": "oral:a-machine-learning-workflow-to-accelerate-the-design-of-in-vitro-release-tests-from-liposomes", "title": "A machine learning workflow to accelerate the design of in vitro release tests from liposomes", "speaker": "Daniel Yanes", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=0b3qTimpmj", "openreview_id": "0b3qTimpmj", "speaker_profile_url": "https://openreview.net/profile?id=~Daniel_Yanes1" } ] }, { "session_title": "Machine Learning for Materials and Interatomic Potentials", "rationale": "When Machine Learning Force Fields Fail Expectations: Lessons We Learned from Solid Electrolyte Materials, Learning Nonlinear Dissolution Trajectories in Binary Polymer–Solvent Systems, Atomic Sudoku: Stochastic approaches for correlated disorder materials, Platonic representation of foundation machine learning interatomic potentials, The Zintl–Klemm Concept in the Amorphous State: A Case Study of Na–P Battery Anodes, Importance of Electronic Entropy for Machine Learning Interatomic Potentials", "section": "AI for Materials Science", "breakout_venue_name": "Sophia", "chair_name": "Wen Jie Ong", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Yizhou Zhu", "title": "When Machine Learning Force Fields Fail Expectations: Lessons We Learned from Solid Electrolyte Materials", "speaker": "Yizhou Zhu", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Machine learning force fields (MLFFs) have emerged as powerful tools for molecular dynamics (MD) simulations of complex materials, offering near-ab initio accuracy at significantly reduced computational cost. In solid electrolyte research, MLFFs enable us to study critical phenomena, including phase transition behaviors and ionic transport mechanisms, providing fundamental insights into these materials.\n\nHowever, a fundamental challenge persists: when MLFF-based MD simulations yield unexpected or even unphysical results, determining whether discrepancies originate from could be difficult due to the black-box nature of MLFFs. Discrepancies could come from limitations of MLFF frameworks, insuffciently sampled training datasets, or overlooked physical mechanisms.\n\nIn this talk, I will share our experience in using MLFF for MD simulations in several solid electrolyte materials. Specifically, we (try to) address the following key questions:\n1) Why our MLFFs work well for cubic-LLZO but always fails in tetragonal-LLZO?\n2) Why do our MLFFs exhibit good training metrics yet still fail during long-time or large-scale production MD simulations?\n3) How much training data is required to develop a reliable MLFF for solid electrolytes, and is highly accurate DFT reference data essential?\n4) What are the trade-offs involved in explicitly including long-range interactions in terms of force field accuracy and performance?\n\nWe hope our successes and failures (in particular failures) of applying MLFFs to MD simulations in solid electrolyte materials may provide valuable insights for the community, offering practical lessons for extending MLFFs to other complex material systems.", "speaker_profile_url": "https://en.westlake.edu.cn/faculty/yizhou-zhu.html", "section": [ "Self-Driving Labs", "AI for Materials Science" ] }, { "talk_id": "oral:learning-nonlinear-dissolution-trajectories-in-binary-polymersolvent-systems", "title": "Learning Nonlinear Dissolution Trajectories in Binary Polymer–Solvent Systems", "speaker": "Zheng Jie Liew", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=DoNF4v49XI", "openreview_id": "DoNF4v49XI", "speaker_profile_url": "https://openreview.net/profile?id=~Liew_Zheng_Jie1" }, { "talk_id": "oral:atomic-sudoku-stochastic-approaches-for-correlated-disorder-materials", "title": "Atomic Sudoku: Stochastic approaches for correlated disorder materials", "speaker": "Andy Paul Chen", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=9zbphxweQ9", "openreview_id": "9zbphxweQ9" }, { "talk_id": "oral:platonic-representation-of-foundation-machine-learning-interatomic-potentials", "title": "Platonic representation of foundation machine learning interatomic potentials", "speaker": "Zhenzhu Li", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=5Yt1eVV5gg", "openreview_id": "5Yt1eVV5gg", "speaker_profile_url": "https://openreview.net/profile?id=~Zhenzhu_Li1" }, { "talk_id": "oral:the-zintlklemm-concept-in-the-amorphous-state-a-case-study-of-nap-battery-anodes", "title": "The Zintl–Klemm Concept in the Amorphous State: A Case Study of Na–P Battery Anodes", "speaker": "Litong Wu", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=5AxyHcmWEZ", "openreview_id": "5AxyHcmWEZ", "speaker_profile_url": "https://openreview.net/profile?id=~Litong_Wu1" }, { "talk_id": "oral:importance-of-electronic-entropy-for-machine-learning-interatomic-potentials", "title": "Importance of Electronic Entropy for Machine Learning Interatomic Potentials", "speaker": "Martin Hoffmann Petersen", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=0DqWkDsL9o", "openreview_id": "0DqWkDsL9o", "speaker_profile_url": "https://openreview.net/profile?id=~Martin_Hoffmann_Petersen1" } ], "chair_profile_url": "https://developer.nvidia.com/blog/author/wong" }, { "session_title": "Autonomous Discovery in Chemistry and Materials", "rationale": "Polymer Discovery through Modular Chemistry Plus Modular Automation, Autonomous nanoparticle synthesis by design, Kinetic study of the aqueous Kolbe-Schmitt reaction enabled by automated reaction analysis, Deciphering the operando mechanism of Haber-Bosch process, A Data-driven Closed-looped High-throughput Platform for Thermocatalyst Discovery, Development of a Platform for Sustainable Metal-Organic Framework (MOF) Synthesis, MABIL: MOF Automation using Biomass-Inspired Linkers", "section": "Self-Driving Labs", "breakout_venue_name": "Moor", "chair_name": "Felix Hanke", "start_time": "14:45", "end_time": "16:15", "talks": [ { "talk_id": "invited:Keith A. Brown", "title": "Polymer Discovery through Modular Chemistry Plus Modular Automation", "speaker": "Keith A. Brown", "talk_type": "featured", "length": 1, "start_time": "14:45", "end_time": "15:00", "abstract": "Accelerating the discovery and development of new polymers is a crucial step towards meeting pressing societal challenges. Data-driven approaches are a powerful part of this process but require high-quality and consistent experimental data spanning a meaningful portion of composition space. Unfortunately, polymers are an especially difficult material class in the context of informatics due to the high importance of processing and large variations in protocols. Further, it is often very difficult to compare polymers that are synthesized using different methods. Here, we report our work to springboard polymer discovery and development through the combination of modular chemistry and modular automation. First, we discuss the electrodeposition of polymer networks (EPoN), a flexible approach to depositing and modifying polymer films. This process involves the attachment of electroactive crosslinking groups onto polymers or prepolymers and subsequently depositing them as conformal and self-limiting films. We have shown this method is capable of realizing a wide array of homopolymer and block copolymer functional films under comparable conditions. To turn this process into a platform for polymer discovery, we combine it with the polymer analysis and discovery array (PANDA), a low-cost robotic system that is able to dispense fluid, perform electrochemistry, and optically characterize EPoN films in custom well plates. The PANDA is part of a larger automated distributed equipment network (DEN) wherein samples can be automatically moved to different stations for structural, analytical, and functional characterization including Raman and UV-Vis spectroscopy, contact angle measurement, and mechanical characterization including adhesion and stiffness. We discuss our initial campaigns to use the combination of EPoN and automation to seek PFAS-replacing non-wetting polymers and optimize materials for carbon capture. Ultimately, the combination of modular chemistry and automation al", "speaker_profile_url": "https://kablab.org/", "section": [ "AI for Materials Science", "Self-Driving Labs", "AI for Physics" ] }, { "talk_id": "oral:autonomous-nanoparticle-synthesis-by-design", "title": "Autonomous nanoparticle synthesis by design", "speaker": "Andy Anker", "talk_type": "contributed", "length": 1, "start_time": "15:00", "end_time": "15:15", "abstract_url": "https://openreview.net/forum?id=KoUjA29US1", "openreview_id": "KoUjA29US1", "speaker_profile_url": "https://openreview.net/profile?id=~Andy_Sode_Anker1" }, { "talk_id": "oral:kinetic-study-of-the-aqueous-kolbe-schmitt-reaction-enabled-by-automated-reaction-analysis", "title": "Kinetic study of the aqueous Kolbe-Schmitt reaction enabled by automated reaction analysis", "speaker": "Muye Xiao", "talk_type": "contributed", "length": 1, "start_time": "15:15", "end_time": "15:30", "abstract_url": "https://openreview.net/forum?id=ud8BSC4AiQ", "openreview_id": "ud8BSC4AiQ", "speaker_profile_url": "https://openreview.net/profile?id=~Muye_Xiao1" }, { "talk_id": "oral:deciphering-the-operando-mechanism-of-haber-bosch-process", "title": "Deciphering the operando mechanism of Haber-Bosch process", "speaker": "Yuanyuan Zhou", "talk_type": "contributed", "length": 1, "start_time": "15:30", "end_time": "15:45", "abstract_url": "https://openreview.net/forum?id=fCvMCCFEMQ", "openreview_id": "fCvMCCFEMQ", "speaker_profile_url": "https://openreview.net/profile?id=~Yuanyuan_Zhou5" }, { "talk_id": "oral:a-data-driven-closed-looped-high-throughput-platform-for-thermocatalyst-discovery", "title": "A Data-driven Closed-looped High-throughput Platform for Thermocatalyst Discovery", "speaker": "Zhengzuo Liu", "talk_type": "contributed", "length": 1, "start_time": "15:45", "end_time": "16:00", "abstract_url": "https://openreview.net/forum?id=TTXJy0GALl", "openreview_id": "TTXJy0GALl", "speaker_profile_url": "https://openreview.net/profile?id=~Zhengzuo_Liu1" }, { "talk_id": "oral:development-of-a-platform-for-sustainable-metal-organic-framework-mof-synthesis-mabil-mof-automation-using-biomass-inspired-linkers", "title": "Development of a Platform for Sustainable Metal-Organic Framework (MOF) Synthesis, MABIL: MOF Automation using Biomass-Inspired Linkers", "speaker": "Ailsa Edward", "talk_type": "contributed", "length": 1, "start_time": "16:00", "end_time": "16:15", "abstract_url": "https://openreview.net/forum?id=GA6vvicz35", "openreview_id": "GA6vvicz35", "speaker_profile_url": "https://openreview.net/profile?id=~Ailsa_K._Edward1" } ], "chair_profile_url": "https://www.cusp.ai/" } ] }, { "block_name": "evening_session", "start_time": "16:45", "end_time": "18:00", "sessions": [ { "session_title": "AI for Materials Discovery and Property Prediction", "rationale": "Simulation at GPU Speed: Reimagining Atomistic Simulation at Scale with NVIDIA ALCHEMI, 🏆 Best Poster: Computational Investigation and Generation of Site-Disordered Sodium Ion Cathode Materials, LeMat-GenBench: A Unified Evaluation Framework for Crystal Generative Models, Using Time-Series Forecasting to Accelerate Materials Stability Assessments, Symmetry‐Aware Equivariant Network for Discovering SHG‐Active Materials", "section": "AI for Materials Science", "breakout_venue_name": "Olivia", "chair_name": "Alex Hammer", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Atul Thakur", "title": "Simulation at GPU Speed: Reimagining Atomistic Simulation at Scale with NVIDIA ALCHEMI", "speaker": "Atul Thakur", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Atomistic simulation is undergoing a fundamental shift, driven by the rapid maturation of machine learning interatomic potentials (MLIPs). Foundation models now deliver near-quantum accuracy across the periodic table, yet the surrounding simulation infrastructure has struggled to keep pace. Most engines remain CPU-centric legacy code, leaving GPUs underutilized and bottlenecked by neighbor list construction, dispersion corrections, electrostatics, and per-system serial execution. \nWe present the NVIDIA ALCHEMI Toolkit — a GPU-first, composable stack designed to accelerate the full MLIP workflow, from training and fine-tuning to production simulation. ALCHEMI Toolkit-Ops provides batched, differentiable primitives (neighbor lists, DFT-D3, PME/Ewald electrostatics, and MD integrators) with PyTorch and JAX bindings that drop into any existing MLIP pipeline. The Toolkit is designed to be easy to use — researchers can compose and scale multi-stage batched pipelines in a few lines of Python. We will demonstrate a few concrete use cases and highlight adoption across verticals and discuss how the community can shape the roadmap.", "section": [ "AI for Chemistry", "AI for Materials Science" ] }, { "talk_id": "extra:keynote:best-poster-in-materials-science:1", "title": "🏆 Best Poster: Computational Investigation and Generation of Site-Disordered Sodium Ion Cathode Materials", "speaker": "Martin Hoffmann Petersen", "talk_type": "extra_keynote", "length": 1, "start_time": "17:00", "end_time": "17:15", "fixed_slot_start": "17:00", "fixed_slot_end": "17:15", "abstract": "Inverse design of sodium-ion cathode materials with machine learning interatomic potentials capable of efficiently and accurately describing various site-disordered phases within the phase space. https://openreview.net/forum?id=xJeKm7k3cL", "speaker_profile_url": "https://openreview.net/profile?id=~Martin_Hoffmann_Petersen1" }, { "talk_id": "oral:lemat-genbench-a-unified-evaluation-framework-for-crystal-generative-models", "title": "LeMat-GenBench: A Unified Evaluation Framework for Crystal Generative Models", "speaker": "Nikita Kazeev", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=my0e7dPuDE", "openreview_id": "my0e7dPuDE" }, { "talk_id": "oral:using-time-series-forecasting-to-accelerate-materials-stability-assessments", "title": "Using Time-Series Forecasting to Accelerate Materials Stability Assessments", "speaker": "Manuel Kober-Czerny", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=7frNW7jaWl", "openreview_id": "7frNW7jaWl", "speaker_profile_url": "https://openreview.net/profile?id=~Manuel_Kober-Czerny1" }, { "talk_id": "oral:symmetryaware-equivariant-network-for-discovering-shgactive-materials", "title": "Symmetry‐Aware Equivariant Network for Discovering SHG‐Active Materials", "speaker": "Ivan Trofimov", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=2H2UuzDQkb", "openreview_id": "2H2UuzDQkb", "speaker_profile_url": "https://openreview.net/profile?id=~Ivan_S._Trofimov1" } ] }, { "session_title": "Autonomous and Self-Driving Laboratory Systems", "rationale": "A digital twin for sim-to-real chemistry lab automation, A Standard Physical Environment for Benchmarking AI-driven Cell Biology, MiST: Understanding the Role of Mid-Stage Scientific Training in Developing Chemical Reasoning Models, Democratizing Discovery: Ultra-Low-Cost Self-Driving Laboratories for Materials Science, Automated High Throughput Optimization for Halide Perovskite Memristors", "section": "Self-Driving Labs", "breakout_venue_name": "Sophia", "chair_name": "Artem Mishchenko", "chair_profile_url": "https://uk.linkedin.com/in/amishche", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Kourosh Darvish", "title": "A digital twin for sim-to-real chemistry lab automation", "speaker": "Kourosh Darvish", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "What if you could design, train, and validate an entire robotic chemistry workflow before running a single real experiment?\n\nThis talk presents Matterix, a GPU-accelerated digital twin framework that captures key physical and operational elements of a chemistry lab, including robotic manipulation, liquids and powders, heat transfer, and known reaction kinetics, within a photorealistic, physics-grounded environment. A modular semantics engine bridges continuous dynamics and discrete logical states, enabling workflow modeling across multiple levels of abstraction.\n\nOn top of this foundation, Matterix enables training of robotic skills and control policies directly in simulation, which can then be transferred to real systems. We demonstrate sim-to-real deployment in chemistry lab automation, reducing reliance on costly experimentation and enabling workflows to be designed, tested, and validated entirely in silico.", "speaker_profile_url": "https://kouroshdarvish.com/", "section": [ "Self-Driving Labs" ] }, { "talk_id": "oral:a-standard-physical-environment-for-benchmarking-ai-driven-cell-biology", "title": "A Standard Physical Environment for Benchmarking AI-driven Cell Biology", "speaker": "Yang Choo", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=pjBLevi2j2", "openreview_id": "pjBLevi2j2", "speaker_profile_url": "https://openreview.net/profile?id=~Yang_Choo1" }, { "talk_id": "oral:mist-understanding-the-role-of-mid-stage-scientific-training-in-developing-chemical-reasoning-models", "title": "MiST: Understanding the Role of Mid-Stage Scientific Training in Developing Chemical Reasoning Models", "speaker": "Tong Xie", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=363e8WyvLm", "openreview_id": "363e8WyvLm", "speaker_profile_url": "https://openreview.net/profile?id=~Tong_Xie2" }, { "talk_id": "oral:democratizing-discovery-ultra-low-cost-self-driving-laboratories-for-materials-science", "title": "Democratizing Discovery: Ultra-Low-Cost Self-Driving Laboratories for Materials Science", "speaker": "Sayan Doloi", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=kmTALke912", "openreview_id": "kmTALke912", "speaker_profile_url": "https://openreview.net/profile?id=~Sayan_Doloi1" }, { "talk_id": "oral:automated-high-throughput-optimization-for-halide-perovskite-memristors", "title": "Automated High Throughput Optimization for Halide Perovskite Memristors", "speaker": "Shreyas Pethe", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=hqG54HVsO9", "openreview_id": "hqG54HVsO9", "speaker_profile_url": "https://openreview.net/profile?id=~Pethe_Shreyas_Dinesh1" } ] }, { "session_title": "AI for proteins, genetic circuits, and cellular biophysics", "rationale": "AFPFusionLM: A Hybrid Sequence–Structure Protein Language Model for Antifreeze Protein Function Prediction, Multi-objective optimization for designing structurally similar proteins with dissimilar sequences, End-to-end neural reconstruction of DNA structures from single-frame fluorescence images, Better Protein Function Prediction by Modeling Survivorship Bias, Risk-averse optimization of genetic circuits under uncertainty", "section": "AI for Biology", "breakout_venue_name": "Moor", "chair_name": "Yongcun Song", "chair_profile_url": "https://dr.ntu.edu.sg/entities/person/Yongcun-Song", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "oral:afpfusionlm-a-hybrid-sequencestructure-protein-language-model-for-antifreeze-protein-function-prediction", "title": "AFPFusionLM: A Hybrid Sequence–Structure Protein Language Model for Antifreeze Protein Function Prediction", "speaker": "Yijun Li", "talk_type": "contributed", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract_url": "https://openreview.net/forum?id=n83shSmu62", "openreview_id": "n83shSmu62", "speaker_profile_url": "https://openreview.net/profile?id=~Yijun_Li9" }, { "talk_id": "oral:multi-objective-optimization-for-designing-structurally-similar-proteins-with-dissimilar-sequences", "title": "Multi-objective optimization for designing structurally similar proteins with dissimilar sequences", "speaker": "Ryo Akiba", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=wXu96N9h54", "openreview_id": "wXu96N9h54", "speaker_profile_url": "https://openreview.net/profile?id=~Ryo_Akiba1" }, { "talk_id": "oral:end-to-end-neural-reconstruction-of-dna-structures-from-single-frame-fluorescence-images", "title": "End-to-end neural reconstruction of DNA structures from single-frame fluorescence images", "speaker": "Fushuai Wang", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=tDe3PzhrQk", "openreview_id": "tDe3PzhrQk", "speaker_profile_url": "https://openreview.net/profile?id=~Fushuai_Wang1" }, { "talk_id": "oral:better-protein-function-prediction-by-modeling-survivorship-bias", "title": "Better Protein Function Prediction by Modeling Survivorship Bias", "speaker": "Poompol Buathong", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=AZwytFG2pI", "openreview_id": "AZwytFG2pI", "speaker_profile_url": "https://openreview.net/profile?id=~Poompol_Buathong1" }, { "talk_id": "oral:risk-averse-optimization-of-genetic-circuits-under-uncertainty", "title": "Risk-averse optimization of genetic circuits under uncertainty", "speaker": "Michal Kobiela", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=4J44HDgitf", "openreview_id": "4J44HDgitf", "speaker_profile_url": "https://openreview.net/profile?id=~Michał_Kobiela1" } ] }, { "session_title": "AI for Quantum Materials and Electronic Structure", "rationale": "Characterizing electronic and structural states of materials with ML, $2k_F$ instability and chiral spin density wave at the 1/9 magnetization plateau in the kagome antiferromagnets, HSG-12M: A Large-Scale Benchmark of Spatial Multigraphs from the Energy Spectra of Non-Hermitian Crystals, Flow-Distorted Plane Waves, Global Plane Waves From Local Gaussians: Periodic Charge Densities in a Blink", "section": "AI for Physics", "breakout_venue_name": "Hullet", "chair_name": "Qianxiao Li", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Simon Billinge", "title": "Characterizing electronic and structural states of materials with ML", "speaker": "Simon Billinge", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "At the heart of materials science studies for next generation materials is an idea that we want to be studying real materials doing real things, often in real devices in real time. In practice, this presents a number of key data analysis and interpretation challenges because it implies we are studying ever more complicated samples, often in complex heterogeneous environments and in time-resolved operando setups, and we are interrogating our data for more and more subtle effects such as microstructures and evolving defects and local structures. Advanced data analysis algorithms and software are essential for the success of this enterprise.  In this talk I will describe various developments that leverage the power of artificial intelligence (AI), principally machine learning (ML), to aid in this task. Some of these powerful tools are clearly ready to be applied more broadly in the community and others are still in the future but look very promising.  They include unsupervised and supervised machine learning approaches, conventional ML and deep neural networks, including generative models, as well as approaches for autonomous time-resolved experimentation. I will lay out various aspects of ML in structure science using examples from our own work.", "section": [ "AI for Materials Science", "AI for Physics" ], "speaker_profile_url": "https://www.cnsi.ucsb.edu/people/faculty/simon-billinge" }, { "talk_id": "oral:2k-f-instability-and-chiral-spin-density-wave-at-the-1-9-magnetization-plateau-in-the-kagome-antiferromagnets", "title": "$2k_F$ instability and chiral spin density wave at the 1/9 magnetization plateau in the kagome antiferromagnets", "speaker": "Tanja Duric", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=nV2GKtfbIP", "openreview_id": "nV2GKtfbIP", "speaker_profile_url": "https://openreview.net/profile?id=~Tanja_Đurić1" }, { "talk_id": "oral:hsg-12m-a-large-scale-benchmark-of-spatial-multigraphs-from-the-energy-spectra-of-non-hermitian-crystals", "title": "HSG-12M: A Large-Scale Benchmark of Spatial Multigraphs from the Energy Spectra of Non-Hermitian Crystals", "speaker": "Xianquan Yan", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=Z6jzmkQ03W", "openreview_id": "Z6jzmkQ03W", "speaker_profile_url": "https://openreview.net/profile?id=~Xianquan_Yan1" }, { "talk_id": "extra:contributed:flow-distorted-plane-waves:1", "title": "Flow-Distorted Plane Waves", "speaker": "Zekun Shi, SEA Garena", "talk_type": "sponsor", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract": "The plane wave basis is widely used in Galerkin approximation, due to its periodicity and computational advantage, where the fast Fourier transform (FFT) can be applied. However, since its spatial resolution is uniform, the number of basis functions required can be excessive for problems with rapidly varying local features. We propose an adaptive basis called flow-distorted plane wave (FDPW), where the bijection of a normalizing flow is used to distort the problem domain, hence achieving adaptive resolution. We apply FDPW to Kohn-Sham density functional theory (DFT) calculations to solid-state systems, demonstrating improved speed and memory usage.", "speaker_profile_url": "https://www.sea.com/products/garena" }, { "talk_id": "oral:global-plane-waves-from-local-gaussians-periodic-charge-densities-in-a-blink", "title": "Global Plane Waves From Local Gaussians: Periodic Charge Densities in a Blink", "speaker": "Jonas Elsborg", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=SlKqLAVsuI", "openreview_id": "SlKqLAVsuI", "speaker_profile_url": "https://openreview.net/profile?id=~Jonas_Elsborg1" } ], "chair_profile_url": "https://blog.nus.edu.sg/qianxiaoli/" }, { "session_title": "LLMs, reasoning, and applied AI across education, finance, and modeling", "rationale": "Modeling and Computation in the Space of Language: Symbolic and LLM-Based Approaches, PieLoT — LLM-driven Toolbox for Theorem Proving Education, BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modeling, Beyond Semantic Similarity: A Two-Phase Non-Parametric Retrieval Workflow for Corporate Credit Underwriting, ESS-MOTIFS: Discovering Rubric-Aligned Motifs for Cohort-Level Essay Assessment", "section": "AI Agents and LLMs for Science", "breakout_venue_name": "Morrison", "chair_name": "Aruhan Rui Shi", "start_time": "16:45", "end_time": "18:00", "talks": [ { "talk_id": "invited:Haizhao Yang", "title": "Modeling and Computation in the Space of Language: Symbolic and LLM-Based Approaches", "speaker": "Haizhao Yang", "talk_type": "featured", "length": 1, "start_time": "16:45", "end_time": "17:00", "abstract": "Scientific modeling and computation traditionally rely on structured mathematics and hand-designed algorithms. In this talk, I propose a new perspective: treating both modeling and computation as processes operating within the space of natural language. I will introduce two complementary approaches that realize this vision. The first uses symbolic learning based on tree structures to generate mathematical expressions, where modeling is performed by constructing symbolic trees and computation is governed by operator rules. The Finite Expression Method (FEX) exemplifies this approach by discovering interpretable, high-accuracy solutions to PDEs and physical systems. The second approach employs large language models (LLMs) for automatic code generation and reasoning to translate scientific problem descriptions into formal mathematical models and executable solvers to solve these problems. As an example, the OptimAI framework demonstrates how multi-agent LLM collaboration enables reliable end-to-end optimization problem modeling and solving. Together, these methods point toward a unified paradigm where symbolic and language models form the foundation for interpretable, scalable scientific discovery and computation.", "speaker_profile_url": "https://haizhaoyang.github.io", "section": [ "AI for Mathematics", "AI for Physics" ] }, { "talk_id": "oral:pielot-llm-driven-toolbox-for-theorem-proving-education", "title": "PieLoT — LLM-driven Toolbox for Theorem Proving Education", "speaker": "Qixiang Zhang", "talk_type": "contributed", "length": 1, "start_time": "17:00", "end_time": "17:15", "abstract_url": "https://openreview.net/forum?id=pIHg6p7N5U", "openreview_id": "pIHg6p7N5U", "speaker_profile_url": "https://openreview.net/profile?id=~Zhang_Qixiang1" }, { "talk_id": "oral:bridge-bootstrapping-text-to-control-time-series-generation-via-multi-agent-iterative-optimization-and-diffusion-modeling", "title": "BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modeling", "speaker": "Viktor Schlegel", "talk_type": "contributed", "length": 1, "start_time": "17:15", "end_time": "17:30", "abstract_url": "https://openreview.net/forum?id=Ux0sDwaefV", "openreview_id": "Ux0sDwaefV", "speaker_profile_url": "https://openreview.net/profile?id=~Viktor_Schlegel1" }, { "talk_id": "oral:beyond-semantic-similarity-a-two-phase-non-parametric-retrieval-workflow-for-corporate-credit-underwriting", "title": "Beyond Semantic Similarity: A Two-Phase Non-Parametric Retrieval Workflow for Corporate Credit Underwriting", "speaker": "Linus Ng", "talk_type": "contributed", "length": 1, "start_time": "17:30", "end_time": "17:45", "abstract_url": "https://openreview.net/forum?id=8N9fUs4ORZ", "openreview_id": "8N9fUs4ORZ", "speaker_profile_url": "https://openreview.net/profile?id=~Linus_Ng_Junjia1" }, { "talk_id": "oral:ess-motifs-discovering-rubric-aligned-motifs-for-cohort-level-essay-assessment", "title": "ESS-MOTIFS: Discovering Rubric-Aligned Motifs for Cohort-Level Essay Assessment", "speaker": "Thi Ngoc Nguyen", "talk_type": "contributed", "length": 1, "start_time": "17:45", "end_time": "18:00", "abstract_url": "https://openreview.net/forum?id=5EKpOp5sJL", "openreview_id": "5EKpOp5sJL", "speaker_profile_url": "https://openreview.net/profile?id=~Ngoc_Thi_Nguyen1" } ], "chair_profile_url": "https://aruhanruishi.com/" } ] } ], "breaks": [ { "name": "Tea Break", "start_time": "10:30", "end_time": "11:00" }, { "name": "Break", "start_time": "12:00", "end_time": "12:15" }, { "name": "Lunch", "start_time": "13:15", "end_time": "14:45" }, { "name": "Tea Break", "start_time": "16:15", "end_time": "16:45" } ], "early_morning_chair_profile_url": "https://www.cnsi.ucsb.edu/people/faculty/simon-billinge" } ], "data_snapshot_at": "2026-06-19T00:04:05+08:00" }