Alert button
Picture for Johannes Heinrich

Johannes Heinrich

Alert button

Deep Reinforcement Learning from Self-Play in Imperfect-Information Games

Add code
Bookmark button
Alert button
Jun 28, 2016
Johannes Heinrich, David Silver

Figure 1 for Deep Reinforcement Learning from Self-Play in Imperfect-Information Games
Figure 2 for Deep Reinforcement Learning from Self-Play in Imperfect-Information Games
Figure 3 for Deep Reinforcement Learning from Self-Play in Imperfect-Information Games
Figure 4 for Deep Reinforcement Learning from Self-Play in Imperfect-Information Games

Many real-world applications can be described as large-scale games of imperfect information. To deal with these challenging domains, prior work has focused on computing Nash equilibria in a handcrafted abstraction of the domain. In this paper we introduce the first scalable end-to-end approach to learning approximate Nash equilibria without prior domain knowledge. Our method combines fictitious self-play with deep reinforcement learning. When applied to Leduc poker, Neural Fictitious Self-Play (NFSP) approached a Nash equilibrium, whereas common reinforcement learning methods diverged. In Limit Texas Holdem, a poker game of real-world scale, NFSP learnt a strategy that approached the performance of state-of-the-art, superhuman algorithms based on significant domain expertise.

* updated version, incorporating conference feedback 
Viaarxiv icon
Github IconAccess Paper or Ask Questions
Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon