Chris Wiggins, Just do the best you can: statistical physics approaches to reinforcement learning

Mon, Dec 13, 2021, 1:15 pm
Location: 
Zoom
Speaker(s): 
Sponsor(s): 
CPBF an NSF PFC

The most celebrated corners of machine learning over the past decades are those successful at predicting - e.g., spam classification, medical diagnoses, or cat faces. But machine learning as actually used in practice is commonly prescriptive rather than predictive: decisions must be made in order to maximize a reward. The misuse of predictive approaches for prescriptive policy needs is as old as multivariate regression itself. Such problems are common in health, commerce, and engineering. These problems broadly fall under the umbrella of reinforcement learning. I will motivate and illustrate some of these applications of reinforcement learning, then show how methods from statistical physics, particularly variational and Monte Carlo methods, can be used to extend and improve modeling approaches.

Andrew Gordus: Untangling the web of behaviors used in spider orb-weaving

Mon, Dec 6, 2021, 12:30 pm

Many innate behaviors are the result of multiple sensorimotor programs that are dynamically coordinated to produce higher-order behaviors such as...

Location: Joseph Henry Room, Jadwin Hall

Chris Wiggins, Just do the best you can: statistical physics approaches to reinforcement learning

Mon, Dec 13, 2021, 1:15 pm

The most celebrated corners of machine learning over the past decades are those successful at predicting - e.g., spam...

Location: Zoom

Madhav Mani: TBD

Mon, Jan 31, 2022, 12:30 pm
Location: Joseph Henry Room, Jadwin Hall

Zvonimir Dogic: TBD

Mon, Feb 7, 2022, 12:30 pm
Location: Joseph Henry Room

Naama Brenner: TBD

Mon, Mar 14, 2022, 12:30 pm

Na Ji: TBD

Mon, Mar 21, 2022, 12:30 pm
Location: Joseph Henry Room

Arnold Mathijssen: TBD

Mon, Mar 28, 2022, 12:30 pm
Location: Joseph Henry Room

Gijsje Koenderink: TBD

Mon, Apr 11, 2022, 12:30 pm
Location: Zoom