Skip to Main content Skip to Navigation
Conference papers

Considering Unseen States as Impossible in Factored Reinforcement Learning

Olga Kozlova Olivier Sigaud Pierre-Henri Wuillemin 1 Christophe Meyer
1 DECISION
LIP6 - Laboratoire d'Informatique de Paris 6
Abstract : The Factored Markov Decision Process (fmdp) framework is a standard representation for sequential decision problems under uncertainty where the state is represented as a collection of random variables. Factored Reinforcement Learning (frl) is an Model-based Reinforcement Learning approach to fmdps where the transition and reward functions of the problem are learned. In this paper, we show how to model in a theoretically well-founded way the problems where some combinations of state variable values may not occur, giving rise to impossible states. Furthermore, we propose a new heuristics that considers as impossible the states that have not been seen so far. We derive an algorithm whose improvement in performance with respect to the standard approach is illustrated through benchmark experiments.
Document type :
Conference papers
Complete list of metadata

https://hal.archives-ouvertes.fr/hal-01296687
Contributor : Lip6 Publications <>
Submitted on : Friday, April 1, 2016 - 2:08:59 PM
Last modification on : Friday, January 8, 2021 - 5:32:06 PM

Links full text

Identifiers

Citation

Olga Kozlova, Olivier Sigaud, Pierre-Henri Wuillemin, Christophe Meyer. Considering Unseen States as Impossible in Factored Reinforcement Learning. European Conference on Machine Learning, Sep 2009, Bled, Slovenia. pp.721-735, ⟨10.1007/978-3-642-04180-8_64⟩. ⟨hal-01296687⟩

Share

Metrics

Record views

90