Selecting the State-Representation in Reinforcement Learning
Abstract
The problem of selecting the right state-representation in a reinforcement learning problem is considered. Several models (functions mapping past observations to a finite set) of the observations are given, and it is known that for at least one of these models the resulting state dynamics are indeed Markovian. Without knowing neither which of the models is the correct one, nor what are the probabilistic characteristics of the resulting MDP, it is required to obtain as much reward as the optimal policy for the correct model (or for the best of the correct models, if there are several). We propose an algorithm that achieves that, with a regret of order T^2/3 where T is the horizon time.
Cite
Text
Maillard et al. "Selecting the State-Representation in Reinforcement Learning." Neural Information Processing Systems, 2011.Markdown
[Maillard et al. "Selecting the State-Representation in Reinforcement Learning." Neural Information Processing Systems, 2011.](https://mlanthology.org/neurips/2011/maillard2011neurips-selecting/)BibTeX
@inproceedings{maillard2011neurips-selecting,
title = {{Selecting the State-Representation in Reinforcement Learning}},
author = {Maillard, Odalric-ambrym and Ryabko, Daniil and Munos, Rémi},
booktitle = {Neural Information Processing Systems},
year = {2011},
pages = {2627-2635},
url = {https://mlanthology.org/neurips/2011/maillard2011neurips-selecting/}
}