Near-optimal Optimistic Reinforcement Learning using Empirical Bernstein Inequalities.
Aristide C. Y. TossouDebabrota BasuChristos DimitrakakisPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- function approximation
- machine learning
- sufficient conditions
- reinforcement learning algorithms
- state space
- theoretical analysis
- learning algorithm
- learning process
- empirical data
- learning classifier systems
- information theoretic
- provably near optimal
- data sets
- mixed integer linear programs
- risk bounds
- optimal control
- monte carlo
- optimal policy
- dynamic programming
- multi agent
- bayesian networks