Title:

Midpoint evaluation for CMA-ES

Subtitle:

Raport Badawczy = Research Report ; RB/59/2013

Creator:

Opara, Karol

Publisher:

Instytut Badań Systemowych. Polska Akademia Nauk ; Systems Research Institute. Polish Academy of Sciences

Place of publishing:

Warszawa

Date issued/created:

2013

Description:

11 pages ; 21 cm ; Bibliography p. 11

Subject and Keywords:

Benchmark ; Cma-es ; Midpoint ; Punkt środkowy

Abstract:

This paper concerns one of the state-of-the art evolutionary algorithms - CMA-ES. It consists of sampling from multivariate normal distribution, whose covariance matrix is claimed to approximate the inverse hessian of the objective function. The midpoint of this distribution should be therefore the best linear unbiased estimator of the optimum. This hypothesis was tested on the BBOB 2013 benchmark set using the standard CMA-ES implementation. Evaluation of the objective function in the midpoint neither improves nor deteriorates the performance of the algorithm. Moreover, it turns out that the standard implementation of CMA-ES is competitive but not as good as the best CMA-ES variants, which took parts in the BBOB 2009 competition.

Relation:

Raport Badawczy = Research Report

Resource type:

Text

Detailed Resource Type:

Report

Source:

RB-2013-59

Language:

eng

Language of abstract:

eng

Rights:

Creative Commons Attribution BY 4.0 license

Terms of use:

Copyright-protected material. [CC BY 4.0] May be used within the scope specified in Creative Commons Attribution BY 4.0 license, full text available at: ; -

Digitizing institution:

Systems Research Institute of the Polish Academy of Sciences

Original in:

Library of Systems Research Institute PAS

Projects co-financed by:

Operational Program Digital Poland, 2014-2020, Measure 2.3: Digital accessibility and usefulness of public sector information; funds from the European Regional Development Fund and national co-financing from the state budget.

Access:

Open