Event time:
Friday, October 25, 2019 - 11:00am
Location:
LOM 107
Speaker:
Nisheeth Vishnoi
Event description:
Abstract:
The principle of maximum entropy states that among all possible ways to write a given point as a convex combination of the domain, pick the one that maximizes the entropy of the probability distribution corresponding to the convex combination. This talk will start by reviewing the computational and structural properties of max-entropy distributions over discrete domains, and then move on to defining and studying them over continuous manifolds with applications to quantum mechanics, semidefinite programming, and statistics. The latter part draws from mathematical physics, in particular from the theory of Lie groups.
Based on joint works with Mohit Singh, Damian Straszak, and Jonathan Leake.
Research Area(s):