- This event has passed.
PICS Colloquium: “Learning about learning by many-body systems”
February 26, 2021 at 2:00 PM - 3:00 PM
Abstract: Many-body systems from soap bubbles to suspensions to polymers learn the drives that push them far from equilibrium. This learning has been detected with thermodynamic properties, such as work absorption and strain. We progress beyond these macroscopic properties that were first defined for equilibrium contexts: We quantify statistical mechanical learning with representation learning, a machine-learning model in which information squeezes through a bottleneck. We identify a structural parallel between representation learning and far-from-equilibrium statistical mechanics. Applying this parallel, we measure four facets of many-body systems’ learning: classification ability, memory capacity, discrimination ability, and novelty detection. Numerical simulations of a classical spin glass illustrate our technique. This toolkit exposes self-organization that eludes detection by thermodynamic measures. Our toolkit more reliably and more precisely detects and quantifies learning by matter.
Nicole Yunger Halpern
ITAMP Postdoctoral Fellow at Harvard University Department of Physics
Dr. Nicole Yunger Halpern is an ITAMP Postdoctoral Fellow at Harvard. She re- envisions 19th-century thermodynamics for 21st-century settings—small, quantum, and far-from-equilibrium contexts—using the mathematical toolkit of quantum information theory. Nicole uses quantum thermodynamics as a new lens through which to view the rest of science: atomic, molecular, and optical physics; condensed matter; chemistry; biophysics; and high-energy physics. She calls her research “quantum steampunk,” after the steampunk genre of art and literature that juxtaposes Victorian settings with futuristic technologies.
Nicole completed her physics PhD in 2018, under John Preskill’s auspices at Caltech. Her dissertation won the Ilya Prigogine Prize for a thermodynamics PhD
thesis. In 2020, she received the International Quantum Technology Emerging Researcher Award from the Institute of Physics. Beginning in fall 2021, she will
be a NIST physicist, a QuICS Fellow at the Joint Institute for Quantum Information and Computer Science (QuICS), an affiliate of the Joint Quantum Institute, and an Adjunct Assistant Professor of Physics and of the Institute for Physical Science and Technology at the University of Maryland. Nicole earned her Masters at the Perimeter Scholars International (PSI) program of the Perimeter Institute for Theoretical Physics and her Bachelors at Dartmouth College, where she graduated as a co-valedictorian of her class.
Nicole writes one story per month for Quantum Frontiers, the blog of Caltech’s Institute for Quantum Information and Matter (https://quantumfrontiers.com/author/nyungerhalpern/). You can connect with her on Twitter @nicoleyh11.Abstract:Many-body systems from soap bubbles to suspensions to polymers learn the drives that push them far from equilibrium. This learning has been detected with thermodynamic properties, such as work absorption and strain. We progress beyond these macroscopic properties that were first defined for equilibrium contexts: We quantify statistical mechanical learning with representation learning, a machine-learning model in which information squeezes through a bottleneck. We identify a structural parallel between representation learning and far-from-equilibrium statistical mechanics. Applying this parallel, we measure four facets of many-body systems’ learning: classification ability, memory capacity, discrimination ability, and novelty detection. Numerical simulations of a classical spin glass illustrate our technique. This toolkit exposes self-organization that eludes detection by thermodynamic measures. Our toolkit more reliably and more precisely detects and quantifies learning by matter.