Safe State Abstraction and Reusable Continuing Subtasks in Hierarchical Reinforcement Learning.
Bernhard HengstPublished in: Australian Conference on Artificial Intelligence (2007)
Keyphrases
- state abstraction
- hierarchical reinforcement learning
- state space
- reinforcement learning
- path finding
- markov decision processes
- initial state
- reinforcement learning algorithms
- path planning
- function approximation
- admissible heuristics
- hierarchical decomposition
- learning algorithm
- dynamic programming
- heuristic search