Learning Transferable Domain Priors for Safe Exploration in Reinforcement Learning.
Thommen George KarimpanalSantu RanaSunil GuptaTruyen TranSvetha VenkateshPublished in: IJCNN (2020)
Keyphrases
- reinforcement learning
- autonomous learning
- learning process
- domain independent
- learning algorithm
- action selection
- learning systems
- supervised learning
- active learning
- prior knowledge
- online learning
- learned knowledge
- active exploration
- exploration exploitation
- reinforcement learning methods
- partially observable
- function approximation
- learning problems
- learning tasks
- complex domains
- learning opportunities
- transfer learning
- domain knowledge
- neural network
- evolutionary learning
- domain specific
- semi supervised