Transfer learning for direct policy search: A reward shaping approach.
Stéphane DoncieuxPublished in: ICDL-EPIROB (2013)
Keyphrases
- transfer learning
- direct policy search
- reward shaping
- reinforcement learning
- learning tasks
- reinforcement learning algorithms
- function approximation
- state space
- learning algorithm
- machine learning
- markov decision processes
- multi agent
- model free
- optimal policy
- mobile robot
- dynamic programming
- decision making
- neural network
- supervised learning
- probabilistic model