Value Iteration under the Constraint of Vector Quantization for Improving Compressed State-action Maps.
Ryuichi UedaTamio AraiPublished in: ICRA (2004)
Keyphrases
- vector quantization
- state action
- average reward
- markov decision process
- markov decision processes
- image compression
- action space
- state space
- vector quantizer
- evaluation function
- reinforcement learning
- optimal policy
- index table
- distortion measure
- belief state
- long run
- dynamic programming
- wavelet transform
- stochastic games
- function approximation
- policy iteration
- model free
- neural network
- image segmentation
- machine learning