Generalized reinforcement learning with concept-driven abstract actions.
Po-Hsiang ChiuManfred HuberPublished in: SMC (2011)
Keyphrases
- reinforcement learning
- action selection
- action space
- perceptual aliasing
- partially observable
- multiagent reinforcement learning
- higher level
- state space
- data driven
- human actions
- model free
- decision theoretic
- state and action spaces
- reasoning about actions
- approximation spaces
- goal directed
- plan recognition
- genetic algorithm
- function approximation
- supervised learning
- dynamic programming
- multi agent
- high level
- learning algorithm