Fall 2016

Learning in Games with Best-Response Oracles

Wednesday, September 21st, 2016 3:30 pm4:00 pm

Add to Calendar


Calvin Lab Auditorium 

What is the computational power of best-response computations in repeated game playing? I will give a precise answer to this question in the context of no-regret learning in zero-sum games, and discuss the implications within the theory of online learning and regret minimization.

Based on joint work with Elad Hazan (STOC'16).