Solving Continuous POMDPs: Value Iteration with Incremental Learning of an Efficient Space Representation.
Sebastian BrechtelTobias GindeleRüdiger DillmannPublished in: ICML (3) (2013)
Keyphrases
- incremental learning
- belief space
- markov decision processes
- partially observable markov decision processes
- incremental learning algorithm
- reinforcement learning
- belief state
- action space
- continuous domains
- state space
- learning process
- supervised learning
- batch learning
- fuzzy artmap
- batch mode
- dynamic programming
- finite state
- data sets
- learning algorithm
- stochastic shortest path
- neural learning
- average reward
- partially observable
- optimal policy
- low dimensional
- markov decision process
- infinite horizon
- heuristic search
- decision theoretic planning
- search space
- e learning