Automated state feature learning for actor-critic reinforcement learning through NEAT.
Yiming PengGang ChenScott HoldawayYi MeiMengjie ZhangPublished in: GECCO (Companion) (2017)
Keyphrases
- reinforcement learning
- actor critic
- temporal difference
- function approximation
- learning process
- learning algorithm
- policy gradient
- state space
- optimal control
- action selection
- learning problems
- reinforcement learning algorithms
- state action
- supervised learning
- model free
- policy iteration
- approximate dynamic programming
- policy gradient methods
- evaluation function
- neuro fuzzy
- partially observable
- temporal difference learning
- rl algorithms
- multi agent
- machine learning
- markov decision processes
- transfer learning
- optimal policy
- function approximators
- average reward
- gradient method
- markov chain
- cost function
- neural network