AlphaExploitem: Going Beyond the Nash Equilibrium in Poker by Learning to Exploit Suboptimal Play
Poker is an imperfect information game that has served as a long-standing benchmark for decision-making under uncertainty. To maximize utility beyond the Nash equilibrium, an agent can deviate from Nash-equilibrium policies to exploit suboptimal play. We introduce AlphaExploitem, which extends the competitive RL poker agent AlphaHoldem by using a hierarchical transformer encoder that enables reasoning over previously played hands and modifying the training procedure with the inclusion of a diverse pool of exploitable opponents to facilitate learning to exploit. We train and evaluate AlphaExploitem on two standard benchmarks for imperfect-information games. Empirically, AlphaExploitem successfully exploits weak play by both in- and out-of-distribution opponents, without losing performance against NE opponents. Poker is a family of card games whose incomplete information setting, stochastic elements, and deep strategic complexity have made it a long-standing challenge for artificial inte
Vlad Murgoci Affiliation: Delft University of Technology Affiliation: 2628 CD Delft, The Netherlands Email: v.murgoci@tudelft.nl Matthijs Spaan Affiliation: Department of Intelligent Systems Affiliation: Delft University of Technology Affiliation: 2628 CD Delft, The Netherlands Email: M.T.J.Spaan@tudelft.nl Yaniv Oren Affiliation: Department of Intelligent Systems Affiliation: Delft University of Technology Affiliation: 2628 CD Delft, The Netherlands Email: y.oren@tudelft.nl Abstract Poker is an imperfect information game that has served as a long-standing benchmark for…
saved by
related reading
- Natural Deception with RL - Rajan Agarwalrajan.sh
- Poker is a bad game for teaching epistemics. Figgie is a better one. — LessWronglesswrong.com
- Noam Brown | Innovators Under 35innovatorsunder35.com
- Reevaluating Policy Gradient Methods for Imperfect-Information Gamesarxiv.org
- Folk theorem (game theory) - Wikipediaen.wikipedia.org
- Nash equilibrium - Wikipediaen.wikipedia.org
- Life is Poker, Not Chess - by Robin Guo - The Bird's Nestrobin-guo.com
- Can AI Learn From Experience? EBR-Bench Results | Epoch AI | Epoch AIepoch.ai
- Learning To Play Settlers of Catan With Deep RLsettlers-rl.github.io
- VGC-Bench: Towards Mastering Diverse Team Strategies in Competitive Pokémonarxiv.org
- Reinforcement Learning in Newcomblike Problemsproceedings.neurips.cc
- Grandmaster-Level Chess Without Searcharxiv.org