Self-Regulating Action Exploration in Reinforcement Learning.
Teck-Hou TengAh-Hwee TanYuan-Sin TanPublished in: INNS-WC (2012)
Keyphrases
- action selection
- reinforcement learning
- active exploration
- action space
- exploration strategy
- temporal difference
- reward shaping
- autonomous learning
- partially observable domains
- reinforcement learning algorithms
- model based reinforcement learning
- state space
- function approximation
- exploration exploitation
- machine learning
- transition model
- markov decision processes
- sensory inputs
- state action
- dynamic programming
- model free
- human actions
- neural network
- fitted q iteration
- reward signal
- balancing exploration and exploitation
- reinforcement learning methods
- reasoning about actions
- markov decision process
- real robot
- mobile robot
- learning process
- learning algorithm