On using discretized Cohen-Grossberg node dynamics for model-free actor-critic neural learning in non-Markovian domains.
Eiji MizutaniStuart E. DreyfusPublished in: CIRA (2003)
Keyphrases
- model free
- actor critic
- reinforcement learning
- neural learning
- temporal difference
- reinforcement learning algorithms
- policy iteration
- multi layered
- function approximation
- average reward
- transfer learning
- neural network
- incremental learning
- policy gradient
- rl algorithms
- state space
- approximate dynamic programming
- dynamical systems
- temporal difference learning
- reinforcement learning methods
- multi agent
- reward function
- markov decision processes
- learning algorithm
- partially observable
- optimal control
- optimal policy
- fixed point
- function approximators
- supervised learning
- dynamic programming