Abstract Demonstrations and Adaptive Exploration for Efficient and Stable Multi-step Sparse Reward Reinforcement Learning.
Xintong YangZe JiJing WuYu-Kun LaiPublished in: CoRR (2022)
Keyphrases
- multi step
- reinforcement learning
- exploration strategy
- distance computation
- lower bounding
- high dimensional
- state space
- reinforcement learning algorithms
- agent receives
- data sets
- function approximation
- learning capabilities
- sparse representation
- single step
- linear combination
- inverse reinforcement learning
- semi supervised
- tumor classification
- eligibility traces
- neural network