Transfer Learning in Reinforcement Learning Problems Through Partial Policy Recycling.
Jan RamonKurt DriessensTom CroonenborghsPublished in: ECML (2007)
Keyphrases
- transfer learning
- reinforcement learning problems
- reinforcement learning
- reinforcement learning algorithms
- learning tasks
- function approximators
- reinforcement learning methods
- markov decision problems
- function approximation
- policy iteration
- model free
- machine learning
- learning algorithm
- temporal difference
- optimal policy
- cross domain
- collaborative filtering
- text classification
- active learning
- labeled data
- markov decision processes
- state space
- text mining
- action space
- transfer knowledge
- machine learning algorithms
- semi supervised learning
- unsupervised learning
- markov decision process
- text categorization
- supervised learning
- optimal control
- long run
- dynamical systems