Timezone: »

 
Poster
Combining Deep Reinforcement Learning and Search for Imperfect-Information Games
Noam Brown · Anton Bakhtin · Adam Lerer · Qucheng Gong

Thu Dec 10 09:00 AM -- 11:00 AM (PST) @ Poster Session 5 #1514
in Poster Session 6 »

The combination of deep reinforcement learning and search at both training and test time is a powerful paradigm that has led to a number of successes in single-agent settings and perfect-information games, best exemplified by AlphaZero. However, prior algorithms of this form cannot cope with imperfect-information games. This paper presents ReBeL, a general framework for self-play reinforcement learning and search that provably converges to a Nash equilibrium in any two-player zero-sum game. In the simpler setting of perfect-information games, ReBeL reduces to an algorithm similar to AlphaZero. Results in two different imperfect-information games show ReBeL converges to an approximate Nash equilibrium. We also show ReBeL achieves superhuman performance in heads-up no-limit Texas hold'em poker, while using far less domain knowledge than any prior poker AI.

Keywords: Algorithms -> Density Estimation; Algorithms -> Similarity and Distance Learning; Applications -> Computer Vision; Theory · Deep Learning

Author Information

Noam Brown (Facebook AI Research)
Anton Bakhtin (Facebook AI Research)
Adam Lerer (Facebook AI Research)
Qucheng Gong (Facebook AI Research)

More from the Same Authors