Smart exploration in reinforcement learning using absolute temporal difference errors.
Clement GehringDoina PrecupPublished in: AAMAS (2013)
Keyphrases
- temporal difference
- reinforcement learning
- action selection
- function approximation
- td learning
- model free
- reinforcement learning algorithms
- temporal difference learning
- evaluation function
- policy evaluation
- function approximators
- step size
- monte carlo
- actor critic
- temporal difference methods
- learning algorithm
- policy iteration
- markov decision processes
- decision making
- transfer learning
- optimal control
- action space
- machine learning
- supervised learning
- dynamic programming
- genetic algorithm