Penalized Proximal Policy Optimization for Safe Reinforcement Learning.
Linrui ZhangLi ShenLong YangShixiang ChenXueqian WangBo YuanDacheng TaoPublished in: IJCAI (2022)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- optimization algorithm
- optimization problems
- markov decision process
- optimization method
- function approximators
- function approximation
- action selection
- partially observable domains
- partially observable environments
- least squares
- markov decision problems
- optimal control
- learning algorithm
- actor critic
- maximum likelihood
- action space
- global optimization
- reinforcement learning methods
- policy gradient methods
- reinforcement learning problems
- state space
- approximate dynamic programming
- policy evaluation
- policy gradient
- continuous state
- reward function
- model free
- finite state
- constrained optimization
- optimization process
- markov decision processes