Policy Improvement for POMDPs Using Normalized Importance Sampling.
Christian R. SheltonPublished in: UAI (2001)
Keyphrases
- importance sampling
- monte carlo
- partially observable markov decision processes
- markov chain
- optimal policy
- kalman filter
- policy gradient
- particle filter
- partially observable
- state space
- variance reduction
- reinforcement learning
- particle filtering
- belief state
- rare events
- approximate inference
- dynamic programming
- infinite horizon
- markov chain monte carlo
- posterior distribution
- distributed systems
- genetic algorithm
- expected reward
- hidden markov models
- evolutionary algorithm
- pairwise
- computer vision