Department of
Applied
Mathematics

Two Worlds Behind Computers Playing Games and How to Bring Them Together (Jirka Matoušek prize talk)

Martin Schmidt

Abstract

During the last few years, we have seen exciting breakthroughs in computer algorithms playing games. Go, Atari Games, Poker - all of these and more are now games in which computers achieve super-human performance. But as it turns out, there are two competing worlds, classes of algorithms and formalisms behind these success stories - the world of reinforcement learning and the world of game theory. First part of this talk will explain what are the differences, and why there historically is quite a large gap between these communities. The second part of the talk will bring some optimism for bringing these two worlds together.