A Surrogate-Assisted Controller for Expensive Evolutionary Reinforcement Learning.
Yuxing WangTiantian ZhangYongzhe ChangBin LiangXueqian WangBo YuanPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- optimal control
- control policy
- control system
- evolutionary computation
- control strategies
- function approximation
- actor critic
- adaptive control
- genetic algorithm
- multi agent reinforcement learning
- control architecture
- state space
- optimal policy
- evolutionary methods
- markov decision processes
- control method
- controller design
- temporal difference
- reinforcement learning algorithms
- evolutionary optimization
- closed loop
- real time
- control strategy
- learning algorithm
- neural network
- control scheme
- function approximators
- computationally expensive
- machine learning
- learning classifier systems
- robot control
- robotic manipulator
- action selection
- nonlinear systems
- model free
- evolutionary algorithm
- pid controller
- control algorithm
- multi agent
- error prone
- learning process
- mathematical model
- supervised learning
- dynamic programming