An Information-Theoretic Optimality Principle for Deep Reinforcement Learning.
Felix LeibfriedJordi Grau-MoyaHaitham Bou-AmmarPublished in: CoRR (2017)
Keyphrases
- information theoretic
- information bottleneck
- reinforcement learning
- mutual information
- information theory
- theoretic framework
- multi modality
- state space
- jensen shannon divergence
- bregman divergences
- information theoretic measures
- minimum description length
- kullback leibler divergence
- entropy measure
- kl divergence
- log likelihood
- markov decision processes
- supervised learning
- image registration
- learning algorithm