Recorded 07 November 2024. Erik Winfree of the California Institute of Technology presents "Equilibrium is inference: lessons from a model of liquid-liquid phase separation" at IPAM's Naturalistic Approaches to Artificial Intelligence Workshop. Abstract: Non-equilibrium behaviors are indisputably essential for living systems. That said, equilibrium behaviors are arguably underappreciated, and it's essential to understand their scope in order to clearly delineate what non-equilibrium dynamics brings to the table. In this talk, we will first review how equilibrium models of neural computation have established the capacity of equilibrium physical systems to perform complex information processing tasks including the representation of complex probability distributions and the inference of conditional probabilities in response to data, and how such systems can be trained. We will then apply these insights to understand the behaviors of programmable molecular systems at equilibrium, including well-mixed chemical reaction networks and liquid-liquid phase separation. Learn more online at: https://www.ipam.ucla.edu/programs/workshops/workshop-iii-naturalistic-approaches-to-artificial-intelligence/?tab=overview