Reducing the Memory Footprint of Temporal Difference Learning over Finitely Many States by Using Case-Based Generalization.
Matt DiltsHéctor Muñoz-AvilaPublished in: ICCBR (2010)
Keyphrases
- memory footprint
- temporal difference learning
- fixed point
- function approximation
- evaluation function
- game playing
- reinforcement learning
- memory usage
- approximate value iteration
- temporal difference
- reinforcement learning algorithms
- markov decision process
- neural network
- monte carlo
- supervised learning
- dynamic programming
- active learning