Near Instance-Optimal PAC Reinforcement Learning for Deterministic MDPs.
Andrea TirinzoniAymen Al MarjaniEmilie KaufmannPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- markov decision processes
- dynamic programming
- state space
- optimal policy
- optimal control
- action sets
- average reward
- learning algorithm
- continuous state spaces
- average cost
- control problems
- policy iteration
- stationary policies
- approximate dynamic programming
- policy search
- initially unknown
- state and action spaces
- reinforcement learning algorithms
- function approximation
- markov decision process
- continuous state and action spaces
- finite horizon
- optimal solution
- decision problems
- deterministic domains
- multi agent
- machine learning
- model free
- state dependent
- reward function
- partially observable
- markov decision problems
- dynamical systems
- linear programming
- factored mdps
- rl algorithms
- temporal difference
- model based reinforcement learning
- factored markov decision processes