Imperial College London

DrPetarKormushev

Faculty of EngineeringDyson School of Design Engineering

Lecturer
 
 
 
//

Contact

 

+44 (0)20 7594 9235p.kormushev Website

 
 
//

Location

 

10-12 Prince's GardensSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@inproceedings{Kormushev:2013,
author = {Kormushev, P and Caldwell, DG},
title = {Comparative Evaluation of Reinforcement Learning with Scalar Rewards and Linear Regression with Multidimensional Feedback},
url = {http://kormushev.com/papers/Kormushev_ECML-PKDD_WS_2013.pdf},
year = {2013}
}

RIS format (EndNote, RefMan)

TY  - CPAPER
AB - This paper presents a comparative evaluation of two learningapproaches. The first approach is a conventional reinforcement learningalgorithm for direct policy search which uses scalar rewards by definition.The second approach is a custom linear regression based algorithm thatuses multidimensional feedback instead of a scalar reward. The two approachesare evaluated in simulation on a common benchmark problem:an aiming task where the goal is to learn the optimal parameters for aimingthat result in hitting as close as possible to a given target. The comparativeevaluation shows that the multidimensional feedback providesa significant advantage over the scalar reward, resulting in an order-ofmagnitudespeed-up of the convergence. A real-world experiment with ahumanoid robot confirms the results from the simulation and highlightsthe importance of multidimensional feedback for fast learning
AU - Kormushev,P
AU - Caldwell,DG
PY - 2013///
TI - Comparative Evaluation of Reinforcement Learning with Scalar Rewards and Linear Regression with Multidimensional Feedback
UR - http://kormushev.com/papers/Kormushev_ECML-PKDD_WS_2013.pdf
UR - http://hdl.handle.net/10044/1/26087
ER -