MPR-RL: Multi-Prior Regularized Reinforcement Learning for Knowledge Transfer.
Quantao YangJohannes A. StorkTodor StoyanovPublished in: IEEE Robotics Autom. Lett. (2022)
Keyphrases
- knowledge transfer
- reinforcement learning
- transfer learning
- function approximation
- previously learned
- cross domain
- transferring knowledge
- learning tasks
- state space
- knowledge sharing
- reinforcement learning algorithms
- markov decision processes
- model free
- optimal policy
- prior knowledge
- machine learning
- function approximators
- temporal difference
- transfer knowledge
- mutual adaptation
- learning algorithm
- rl algorithms
- optimal control
- supervised learning
- markov decision process
- information retrieval
- artificial intelligence
- social networks
- reinforcement learning methods
- e learning
- action space
- training data
- software engineering
- action selection
- collaborative filtering
- multi label
- multi task