Learning to reach by reinforcement learning using a receptive field based function approximation approach with continuous actions.
Minija TamosiunaiteTamim AsfourFlorentin WörgötterPublished in: Biol. Cybern. (2009)
Keyphrases
- reinforcement learning
- function approximation
- learning tasks
- temporal difference learning
- function approximators
- temporal difference methods
- learning algorithm
- temporal difference
- action space
- receptive fields
- model free
- td learning
- policy gradient
- action selection
- learning process
- actor critic
- supervised learning
- reinforcement learning algorithms
- state space
- pattern recognition
- reinforcement learning methods
- machine learning
- natural images
- learning agent
- data sets
- reward function
- multiscale
- genetic algorithm