Q-learning in continuous state-action space with redundant dimensions by using a selective desensitization neural network.
Takaaki KobayashiTakeshi ShibuyaMasahiko MoritaPublished in: SCIS&ISIS (2014)
Keyphrases
- state action space
- function approximation
- reinforcement learning
- neural network
- radial basis function
- actor critic
- optimal policy
- function approximators
- artificial neural networks
- state space
- model free
- learning tasks
- temporal difference learning
- reinforcement learning algorithms
- temporal difference
- action space
- action selection
- control problems
- fuzzy logic
- markov decision processes
- learning algorithm
- learning capabilities
- policy gradient
- dynamic programming
- machine learning