Reinforcement learning in partially observable mobile robot domains using unsupervised event extraction.
Bram BakkerFredrik LinåkerJürgen SchmidhuberPublished in: IROS (2002)
Keyphrases
- partially observable
- event extraction
- reinforcement learning
- mobile robot
- partially observable domains
- state space
- markov decision processes
- partial observability
- open domain
- information extraction
- dynamical systems
- online news
- decision problems
- transfer learning
- partially observable environments
- markov decision problems
- supervised learning
- information fusion
- infinite horizon
- cross lingual
- machine learning
- news articles
- action models
- reward function
- belief state
- optimal policy
- function approximation
- partially observable markov decision processes
- optimal control
- model free
- information access
- heuristic search
- learning algorithm
- markov chain