Towards Characterizing Divergence in Deep Q-Learning.
Joshua AchiamEthan KnightPieter AbbeelPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- function approximation
- state space
- learning algorithm
- cooperative
- action selection
- multi agent
- learning rate
- stochastic approximation
- reinforcement learning algorithms
- model free
- data sets
- bucket brigade
- database
- optimal policy
- sufficient conditions
- multi agent reinforcement learning
- fixed point
- markov decision processes
- information theoretic
- dynamic programming
- deep learning
- kl divergence
- relative entropy
- neural network
- potential field
- hierarchical reinforcement learning