A Unified Approach to Reinforcement Learning, Quantal Response Equilibria, and Two-Player Zero-Sum Games.
Samuel SokotaRyan D'OrazioJ. Zico KolterNicolas LoizouMarc LanctotIoannis MitliagkasNoam BrownChristian KroerPublished in: ICLR (2023)
Keyphrases
- reinforcement learning
- game theoretic
- imperfect information
- reinforcement learning algorithms
- nash equilibria
- perfect information
- nash equilibrium
- optimal strategy
- game theory
- repeated games
- decision problems
- function approximation
- learning agents
- fixed point
- opponent modeling
- markov decision processes
- model free
- optimal policy
- pure strategy
- state space
- multi agent
- multiagent learning
- learning algorithm
- temporal difference learning
- temporal difference
- single agent
- multi agent reinforcement learning
- incomplete information
- robotic control
- average reward
- policy iteration
- learning capabilities
- machine learning
- reinforcement learning methods
- solution concepts
- action selection
- optimal control
- worst case