AI4OPT Seminar Series

Date: Thursday, Sept 14, 2023

Time: Noon – 1:00 pm

Location: 9th floor Atrium in Coda Building (756 W Peachtree St NW, Atlanta, GA 30308)

Join Virtual: https://gatech.zoom.us/j/99381428980

Speaker: Akshay Krishnamurthym

Learning Hidden Markov Models Using Conditional Samples

Abstract: Hidden Markov models (HMMs) are some of the most widely-used tools in statistical sequence modeling; unfortunately HMMs are known to be cryptographically hard to learn in the standard setting where one has access to iid samples of observation sequences. In this talk, I will discuss an alternative learning protocol that can circumvent these computational barriers. Specifically, I will present provably efficient algorithms for learning HMMs when given interactive access to samples from the conditional distributions of the HMM. The algorithms can be seen as generalizations and robustifications of Angluin's classic L-star algorithm for learning deterministic finite automata.

Bio: Akshay Krishnamurthy is a principal research manager at Microsoft Research, New York City. His research interests include machine learning and statistics, with a particular focus on the foundations of interactive learning, contextual bandits, and reinforcement learning.

Lunch will be served at the seminar. So, please stop by 15 minutes before the seminar to pick up lunch.

To receive AI4OPT seminar announcements, please sign up to our mailing list. at https://lists.isye.gatech.edu/mailman/listinfo/ai4opt-seminars.

Past seminars can be found at https://www.ai4opt.org/seminars/past-seminars.