BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Fabian Ruehle (Northeastern University) DTSTART:20251204T190000Z DTEND:20251204T200000Z DTSTAMP:20260422T225846Z UID:ai-for-science/1 DESCRIPTION:Title: Symbolic Regression\, Sparsification\, and Kolmogorov-Arnold Networ ks\nby Fabian Ruehle (Northeastern University) as part of FirstPrincip les Talks\n\n\nAbstract\nInterpreting neural networks remains challenging\ , largely due to their dense parametrization\, global coupling of paramete rs\, and the polysemantic behavior of neurons. These problems are ameliora ted in Kolmogorov-Arnold Networks\, which have fewer parameters overall\, parameter changes are contained to local regions\, and there are less poly semantic neurons. \n\nIn the first part of this talk\, Fabian will show ho w KANs can be viewed as neural networks that have undergone a principled s parsification\, clarifying why they exhibit improved interpretability and parameter efficiency. He will then present a new framework for multivariat e symbolic regression that couples KANs\, LLMs\, and genetic search strate gies\, akin to FunSearch\, to discover compact analytic expressions from d ata. This approach enables scalable symbolic regression in high-dimensiona l settings\, leverages the inductive biases inherent in KANs\, and the abi lity to prime the LLM's regression proposals for different data domains.\n LOCATION:https://researchseminars.org/talk/ai-for-science/1/ END:VEVENT BEGIN:VEVENT SUMMARY:Juan Felipe Carrasquilla Alvarez (ETH Zurich) DTSTART:20251211T150000Z DTEND:20251211T160000Z DTSTAMP:20260422T225846Z UID:ai-for-science/2 DESCRIPTION:Title: Language Models for the Simulation of Quantum Many-Body\nby Jua n Felipe Carrasquilla Alvarez (ETH Zurich) as part of FirstPrinciples Talk s\n\n\nAbstract\nIn this talk\, Juan will discuss his work on using models inspired by natural language processing in the realm of quantum many-body physics. He will demonstrate their utility in solving ground states of qu antum Hamiltonians\, particularly for ground states of arrays of Rydberg a toms on the Kagome lattice. The findings highlight the potential of using language models to explore many-body physics on exotic lattices and beyond .\n LOCATION:https://researchseminars.org/talk/ai-for-science/2/ END:VEVENT BEGIN:VEVENT SUMMARY:Jianke Yang (UC San Diego) DTSTART:20251218T190000Z DTEND:20251218T200000Z DTSTAMP:20260422T225846Z UID:ai-for-science/3 DESCRIPTION:Title: Towards AI Co-Scientist: Automatic Governing Law Discovery\nby Jianke Yang (UC San Diego) as part of FirstPrinciples Talks\n\n\nAbstract\ nScientific law discovery has historically been limited by human reasoning and data scarcity\, despite the vast search space of possible formulation s. Advances in generative AI and abundant physical data now enable AI mode ls to extract interpretable structures\, such as symmetries\, differential equations\, and conserved quantities\, and use them as inductive biases i n predictive and generative tasks.\n\n\nJianke's research aims to develop an AI co-scientist\, a unified system that can (1) autonomously discover g overning structures from raw observations\, (2) translate these discoverie s into flexible inductive biases to improve downstream tasks\, and (3) orc hestrate modular tools under a top-level planner to generate hypotheses\, implement models that satisfy physical constraints\, and complete the pipe line from data→law→model→prediction.\n\n\nThis thesis proposal prese nts the following milestones toward this goal. First\, we formulate the pr oblem of symmetry discovery and introduce two models\, LieGAN and LaLiGAN\ , that discover invariance and equivariance from data using a generative a dversarial framework. Second\, we incorporate symmetry into the task of go verning equation discovery\, showing that symmetry is a powerful inductive bias in the discovery of other physical laws. Together\, these serve as t he building blocks towards a fully functional AI co-scientist system.\n LOCATION:https://researchseminars.org/talk/ai-for-science/3/ END:VEVENT BEGIN:VEVENT SUMMARY:Gerard Milburn (National Quantum Computing Centre) DTSTART:20260108T160000Z DTEND:20260108T170000Z DTSTAMP:20260422T225846Z UID:ai-for-science/4 DESCRIPTION:Title: Quantum machines learning quantum\nby Gerard Milburn (National Quantum Computing Centre) as part of FirstPrinciples Talks\n\nAbstract: TB A\n LOCATION:https://researchseminars.org/talk/ai-for-science/4/ END:VEVENT BEGIN:VEVENT SUMMARY:Nathan Kutz (University of Washington) DTSTART:20260113T200000Z DTEND:20260113T210000Z DTSTAMP:20260422T225846Z UID:ai-for-science/5 DESCRIPTION:Title: Shallow Recurrent Decoders for the Automated Discovery of Physical Models\nby Nathan Kutz (University of Washington) as part of FirstPrin ciples Talks\n\nAbstract: TBA\n LOCATION:https://researchseminars.org/talk/ai-for-science/5/ END:VEVENT BEGIN:VEVENT SUMMARY:Chenhao Tan (University of Chicago) DTSTART:20260116T203000Z DTEND:20260116T213000Z DTSTAMP:20260422T225846Z UID:ai-for-science/6 DESCRIPTION:Title: Science in the Age of AI\nby Chenhao Tan (University of Chicago ) as part of FirstPrinciples Talks\n\nAbstract: TBA\n LOCATION:https://researchseminars.org/talk/ai-for-science/6/ END:VEVENT BEGIN:VEVENT SUMMARY:Alexei Koulakov (Cold Spring Harbor Laboratory) DTSTART:20260128T180000Z DTEND:20260128T190000Z DTSTAMP:20260422T225846Z UID:ai-for-science/7 DESCRIPTION:Title: From Neurons to Newtons: What can the brain teach us about physics? \nby Alexei Koulakov (Cold Spring Harbor Laboratory) as part of FirstP rinciples Talks\n\n\nAbstract\nModern physics has been extraordinarily suc cessful at describing the natural world\, yet the process by which new phy sical theories are constructed remains largely artisanal. In this talk\, A lexei will discuss the principles of brain function and evolution which ca n offer tools for building new physics theories. \n\nFirst\, he will intro duce the concept of a genomic bottleneck\, the idea that neural systems ar e forced to compress vast sensory experience into representations that are simple\, robust\, and reusable across tasks. I suggest that similar bottl enecks may be essential for identifying abstractions that generalize acros s subfields of physics. Second\, he will discuss how brains appear to cons truct internal imagination modules\, generative models that allow organism s to simulate physical phenomena and test hypotheses without direct intera ction with the world. Finally\, Alexei will show how hierarchical reinforc ement learning can provide a natural framework for organizing physical rea soning across scales\, from low-level dynamics to high-level concepts. \n\ nBy decomposing complex problems into nested objectives\, hierarchical con trol offers a computational model for how intelligent systems\, biological or artificial\, can efficiently explore and solve hard physics problems. These ideas suggest a neuroscience-inspired roadmap for transforming theor y building in physics: one that emphasizes distillation\, imagination\, an d hierarchical control as core computational primitives.\n LOCATION:https://researchseminars.org/talk/ai-for-science/7/ END:VEVENT BEGIN:VEVENT SUMMARY:Randy Davilla (Rice University\; FirstPrinciples) DTSTART:20260331T150000Z DTEND:20260331T160000Z DTSTAMP:20260422T225846Z UID:ai-for-science/8 DESCRIPTION:Title: Automated Conjecturing\, Internal Theories\, and AI Discovery\n by Randy Davilla (Rice University\; FirstPrinciples) as part of FirstPrinc iples Talks\n\n\nAbstract\nAutomated conjecturing is the computer-assisted generation of human-readable mathematical statements—such as bounds and structural patterns—that experts can test\, refine\, or prove. Building on earlier systems like Graffiti and TxGraffiti\, this approach focuses o n producing interpretable conjectures rather than predictive models.\n\nIn this talk\, we introduce Graffiti3\, a framework that organizes mathemati cal evidence into evolving tables of objects and invariants\, allowing can didate conjectures to be generated\, tested\, and refined through countere xample search and expert review. The system combines deterministic conject ure engines with large language models to propose features and summarize r esults\, illustrating how AI systems can assist in mathematical discovery across domains such as graph theory\, finite groups\, knot theory\, and Ca labi–Yau geometry.\n LOCATION:https://researchseminars.org/talk/ai-for-science/8/ END:VEVENT BEGIN:VEVENT SUMMARY:Philip Harris (MIT) DTSTART:20260416T180000Z DTEND:20260416T190000Z DTSTAMP:20260422T225846Z UID:ai-for-science/9 DESCRIPTION:Title: Around the Forces in 80 Microseconds\nby Philip Harris (MIT) as part of FirstPrinciples Talks\n\n\nAbstract\nWith large amounts of data\, a Higgs boson discovery\, and world-leading constraints on fundamental fo rces\, the Large Hadron Collider has been a phenomenal tool. However\, it is going through a midlife crisis. More data\, more Higgs bosons\, and mor e constraints are not generating the same excitement as in the past. We ve nture into a new direction with fresh insights that enable unprecedented p hysics measurements\, and we ask how we can automate the discovery and mea surement process using artificial intelligence. We then look at the future of the LHC and present a real-time system\, a custom electronics system\, built around similar novel AI-based processing technology that will expan d the scope of future physics measurements at the LHC. We extend the same real-time AI approaches into gravitational wave astrophysics\, highlightin g new results with an end-to-end AI pipeline. Finally\, we comment on a ne w paradigm for next-generation experimental physics research.\n LOCATION:https://researchseminars.org/talk/ai-for-science/9/ END:VEVENT BEGIN:VEVENT SUMMARY:Alfredo Guevera DTSTART:20260428T180000Z DTEND:20260428T190000Z DTSTAMP:20260422T225846Z UID:ai-for-science/10 DESCRIPTION:Title: Gravitons in the age of AI\nby Alfredo Guevera as part of Firs tPrinciples Talks\n\n\nAbstract\nWe showcase a dialog with recent language models that helped tackle novel results in quantum field theory and quant um gravity. We comment on the limitations and prospects of this approach.\ n LOCATION:https://researchseminars.org/talk/ai-for-science/10/ END:VEVENT END:VCALENDAR