Agent's actions as a classification criteria for the state space in a learning from rewards system.
Francisco Martinez-GilPublished in: J. Exp. Theor. Artif. Intell. (2008)
Keyphrases
- reinforcement learning
- state space
- learning agent
- stochastic domains
- partially observable
- supervised learning
- learning algorithm
- action selection
- decision theoretic
- goal directed
- reward function
- macro actions
- autonomous agents
- classification accuracy
- support vector
- markov decision processes
- learning process
- partial observations
- multi agent systems
- simulated robot
- learned knowledge
- learning mechanism
- multiple agents
- multiagent systems
- learning mechanisms
- cognitive agents
- optimal policy
- partially observable markov decision process
- state abstraction
- plan recognition
- training data