Relative Entropy Regularized Policy Iteration.
Abbas AbdolmalekiJost Tobias SpringenbergJonas DegraveSteven BohezYuval TassaDan BelovNicolas HeessMartin A. RiedmillerPublished in: CoRR (2018)
Keyphrases
- relative entropy
- policy iteration
- bregman divergences
- least squares
- markov decision processes
- theoretical guarantees
- model free
- information theoretic
- fixed point
- optimal policy
- reinforcement learning
- covariance matrix
- information theory
- log likelihood
- mahalanobis distance
- mutual information
- infinite horizon
- temporal difference
- finite state
- cost sensitive
- markov decision process
- learning theory
- objective function
- maximum entropy
- exponential family
- linear programming
- state space
- loss function
- kl divergence
- convergence rate
- nearest neighbor
- long run
- optimal control
- function approximation
- boosting algorithms
- pairwise