Hybrid Value Iteration for POMDPs.
Diego ManiloffPiotr J. GmytrasiewiczPublished in: FLAIRS Conference (2011)
Keyphrases
- partially observable markov decision processes
- markov decision processes
- belief state
- state space
- belief space
- dynamic programming
- optimal policy
- partially observable markov
- reinforcement learning
- average reward
- finite state
- infinite horizon
- heuristic search
- partially observable
- information systems
- decision problems
- continuous state
- markov decision problems
- decision making
- planning under uncertainty
- markov decision process
- neural network
- decision processes
- knowledge base
- policy iteration
- search space
- probability distribution
- hybrid approaches
- markov decision chains