# Learning via Non-Convex Min-Max Games

When:
Friday, November 15, 2019, 11:00 am - 12:00 pm PDTiCal
Where:
1016
This event is open to the public.
Type:
AI Seminar
Speaker:
Meisam Razaviyayn, USC
Video Recording:
While the min-max optimization problem has been extensively studied in the convex-concave regime, our knowledge for the non-convex scenario is still very limited. In this talk, we study the problem in the non-convex regime and show that an $\epsilon$--first order stationary point of the game can be computed  when one of the player’s objective can be optimized to global optimality efficiently.  We discuss the application of the proposed algorithm and theory in defense agains adversarial attacks to neural networks, generative adversarial networks, fair learning, and generative adversarial imitation learning.