Experimental study of the eligibility traces in complex valued reinforcement learning.
Takeshi ShibuyaShingo ShimadaTomoki HamagamiPublished in: SMC (2007)
Keyphrases
- experimental study
- eligibility traces
- complex valued
- reinforcement learning
- reinforcement learning algorithms
- real valued
- recurrent neural networks
- policy evaluation
- reinforcement learning methods
- function approximation
- linear equations
- model free
- multi valued
- state space
- temporal difference
- markov decision processes
- learning speed
- control problems
- activation function
- supervised learning
- learning process
- machine learning
- action selection
- neural network
- optimal control
- action space
- transfer learning
- optimal policy