Monte Carlo Augmented Actor-Critic for Sparse Reward Deep Reinforcement Learning from Suboptimal Demonstrations.
Albert WilcoxAshwin BalakrishnaJules DedieuWyame BenslimaneDaniel S. BrownKen GoldbergPublished in: NeurIPS (2022)
Keyphrases
- monte carlo
- actor critic
- reinforcement learning
- policy gradient
- temporal difference
- variance reduction
- average reward
- reinforcement learning algorithms
- markov chain
- function approximation
- importance sampling
- state space
- optimal control
- policy iteration
- particle filter
- approximate dynamic programming
- learning algorithm
- model free
- markov decision processes
- temporal difference learning
- dynamic programming
- multi agent
- optimal policy
- control problems
- action selection
- gradient method
- policy evaluation
- partially observable
- reward function
- game tree
- control policy
- learning problems
- transfer learning
- sparse representation
- search algorithm