Proto-value functions: A Laplacian framework for learning representation and control in Markov decision processes. (2007)

by KALYANAM WU, GIVAN Mahadevan, S, M Maggioni
Venue:Journal of Machine Learning Research,