Independent Policy Gradient for Large-Scale Markov Potential Games: Sharper Rates, Function Approximation, and Game-Agnostic Convergence.
Dongsheng DingChen-Yu WeiKaiqing ZhangMihailo R. JovanovicPublished in: ICML (2022)
Keyphrases
- function approximation
- policy gradient
- game theory
- reinforcement learning
- video games
- game playing
- game play
- nash equilibrium
- nash equilibria
- game theoretic
- learning tasks
- temporal difference
- reinforcement learning algorithms
- stochastic games
- actor critic
- function approximators
- radial basis function
- imperfect information
- temporal difference learning
- policy search
- markov chain
- model free
- convergence speed
- learning experience
- dynamic environments
- training data
- e learning