Reinforcement-learning agents with different temperature parameters explain the variety of human action-selection behavior in a Markov decision process task.
Fumihiko IshidaTakahiro SasakiYutaka SakaguchiHiroyuki ShimaiPublished in: Neurocomputing (2009)
Keyphrases
- action selection
- reinforcement learning agents
- reinforcement learning
- human robot
- basal ganglia
- robot soccer
- multi agent environments
- action selection mechanism
- optimal policy
- multi agent
- dynamic environments
- temporal difference
- feature selection
- markov decision processes
- transition probabilities
- human robot interaction
- simulated annealing
- state abstraction
- mobile robot
- decision making
- learning algorithm