Learning the goal-structure of actions in a connectionist network without inverse planning.
Robert PowersDavid C. PlautPublished in: CogSci (2018)
Keyphrases
- connectionist networks
- macro actions
- stochastic domains
- learning process
- goal directed
- online learning
- action selection
- learning systems
- learning algorithm
- decision theoretic
- macro operators
- learning tasks
- dynamical systems
- partially observable
- explanation based learning
- knowledge acquisition
- supervised learning
- dynamic programming
- active learning
- prior knowledge
- reinforcement learning