value-aware loss function for model learning in ... · that minimizes a probabilistic loss, such as...

MITSUBISHI ELECTRIC RESEARCH LABORATORIES http://www.merl.com Value-Aware Loss Function for Model Learning in Reinforcement Learning Farahmand, A.-M.; Barreto, A.M.S.; Nikovski, D.N. TR2016-153 December 2016 Abstract We consider the problem of estimating the transition probability kernel to be used by a model- based reinforcement learning (RL) algorithm. We argue that estimating a generative model that minimizes a probabilistic loss, such as the log-loss, might be an overkill because such a probabilistic loss does not take into account the underlying structure of the decision problem and the RL algorithm that intends to solve it. We introduce a loss function that takes the structure of the value function into account. We provide a ﬁnite-sample upper bound for the loss function showing the dependence of the error on model approximation error and the number of samples. European Workshop on Reinforcement Learning (EWRL) This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonproﬁt educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved. Copyright c Mitsubishi Electric Research Laboratories, Inc., 2016 201 Broadway, Cambridge, Massachusetts 02139

Upload: others

Post on 03-Sep-2020

3 views

Category:

Documents

0 download

Report

Download

Embed Size (px):

TRANSCRIPT

Page 1: Value-Aware Loss Function for Model Learning in ... · that minimizes a probabilistic loss, such as the log-loss, might be an overkill because such a probabilistic loss does not take

MITSUBISHI ELECTRIC RESEARCH LABORATORIEShttp://www.merl.com

Value-Aware Loss Function for Model Learning inReinforcement Learning

Farahmand, A.-M.; Barreto, A.M.S.; Nikovski, D.N.

TR2016-153 December 2016

AbstractWe consider the problem of estimating the transition probability kernel to be used by a model-based reinforcement learning (RL) algorithm. We argue that estimating a generative modelthat minimizes a probabilistic loss, such as the log-loss, might be an overkill because such aprobabilistic loss does not take into account the underlying structure of the decision problemand the RL algorithm that intends to solve it. We introduce a loss function that takes thestructure of the value function into account. We provide a finite-sample upper bound forthe loss function showing the dependence of the error on model approximation error and thenumber of samples.

European Workshop on Reinforcement Learning (EWRL)

This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy inwhole or in part without payment of fee is granted for nonprofit educational and research purposes provided that allsuch whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi ElectricResearch Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and allapplicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall requirea license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved.