Selecting Near-Optimal Approximate State Representations in Reinforcement Learning.
Ronald OrtnerOdalric-Ambrym MaillardDaniil RyabkoPublished in: ALT (2014)
Keyphrases
- reinforcement learning
- function approximation
- policy evaluation
- relational reinforcement learning
- multi agent reinforcement learning
- model free
- state space
- optimal policy
- databases
- learning problems
- temporal difference
- robot control
- provably near optimal
- robotic control
- learning agents
- markov decision process
- reinforcement learning algorithms
- action selection
- exact solution
- monte carlo
- supervised learning
- search algorithm
- learning algorithm