Risk-Sensitive Soft Actor-Critic for Robust Deep Reinforcement Learning under Distribution Shifts.
Tobias EndersJames HarrisonMaximilian SchifferPublished in: CoRR (2024)
Keyphrases
- actor critic
- risk sensitive
- optimal control
- reinforcement learning
- model free
- markov decision processes
- policy gradient
- reinforcement learning algorithms
- temporal difference
- function approximation
- policy iteration
- dynamic programming
- markov decision problems
- approximate dynamic programming
- neuro fuzzy
- average reward
- learning algorithm
- control policies
- gradient method
- probability distribution
- infinite horizon
- state space
- control strategies
- transition probabilities
- search space
- learning problems
- optimal policy
- multi agent