1-recall reinforcement learning leading to an optimal equilibrium in potential games with discrete and continuous actions.
Tatiana TatarenkoPublished in: CDC (2015)
Keyphrases
- reinforcement learning
- action sets
- continuous action
- action space
- continuous state spaces
- policy search
- continuous state
- continuous state and action spaces
- repeated games
- nash equilibrium
- dynamic programming
- state space
- game theory
- continuous domains
- action selection
- markov decision processes
- nash equilibria
- piecewise linear
- optimal control
- fictitious play
- machine learning
- initially unknown
- worst case
- learning agents
- state and action spaces
- discrete space
- continuous functions
- continuous data
- function approximation
- computer games
- multi agent
- state action
- high precision
- temporal difference
- game theoretic
- reward function
- optimal solution
- pure strategy
- mixed strategy
- average reward
- partially observable
- stochastic games
- discrete data