Hybrid least-squares algorithms for approximate policy evaluation.

Jeffrey Johns Marek Petrik Sridhar Mahadevan

Published in: Mach. Learn. (2009)

Keyphrases

policy evaluation
least squares
policy iteration
hybrid algorithms
reinforcement learning
temporal difference
model free
matrix inversion
monte carlo
linear model
markov decision processes
variance reduction
computer vision
linear regression
state space
approximate solutions