Towards Safe Reinforcement Learning with a Safety Editor Policy.
Haonan YuWei XuHaichao ZhangPublished in: NeurIPS (2022)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- action selection
- partially observable environments
- markov decision processes
- function approximation
- function approximators
- reinforcement learning problems
- actor critic
- policy evaluation
- reinforcement learning algorithms
- action space
- policy iteration
- markov decision problems
- control policy
- control policies
- temporal difference
- reward function
- approximate dynamic programming
- policy gradient
- partially observable markov decision processes
- state action
- state space
- model free
- state dependent
- reinforcement learning methods
- continuous state spaces
- machine learning
- infinite horizon
- average reward
- model free reinforcement learning
- learning algorithm
- partially observable
- policy making
- optimal control
- partially observable domains
- decision problems
- transfer learning
- approximate policy iteration
- inverse reinforcement learning
- state and action spaces
- supervised learning
- policy gradient methods
- neural network
- multi agent systems
- learning problems
- finite state
- policy makers