Convergence Analysis of the Hessian Estimation Evolution Strategy

Research output: Contribution to journal › Journal article › Research › peer-review

Standard

Convergence Analysis of the Hessian Estimation Evolution Strategy. / Glasmachers, Tobias; Krause, Oswin.

In: Evolutionary Computation, Vol. 30, No. 1, 2022, p. 27-50.

Research output: Contribution to journal › Journal article › Research › peer-review

Harvard

Glasmachers, T & Krause, O 2022, 'Convergence Analysis of the Hessian Estimation Evolution Strategy', Evolutionary Computation, vol. 30, no. 1, pp. 27-50. https://doi.org/10.1162/evco_a_00295

APA

Glasmachers, T., & Krause, O. (2022). Convergence Analysis of the Hessian Estimation Evolution Strategy. Evolutionary Computation, 30(1), 27-50. https://doi.org/10.1162/evco_a_00295

Vancouver

Glasmachers T, Krause O. Convergence Analysis of the Hessian Estimation Evolution Strategy. Evolutionary Computation. 2022;30(1):27-50. https://doi.org/10.1162/evco_a_00295

Author

Glasmachers, Tobias ; Krause, Oswin. / Convergence Analysis of the Hessian Estimation Evolution Strategy. In: Evolutionary Computation. 2022 ; Vol. 30, No. 1. pp. 27-50.

Bibtex

@article{b0ef89c54f0a4a19a1bbb2be90d43fef,
title = "Convergence Analysis of the Hessian Estimation Evolution Strategy",
abstract = "The class of algorithms called Hessian Estimation Evolution Strategies (HE-ESs) update the covariance matrix of their sampling distribution by directly estimating the curvature of the objective function. The approach is practically efficient, as attested by respectable performance on the BBOB testbed, even on rather irregular functions. In this article, we formally prove two strong guarantees for the (1 + 4)-HE-ES, a minimal elitist member of the family: stability of the covariance matrix update, and as a consequence, linear convergence on all convex quadratic problems at a rate that is independent of the problem instance.",
keywords = "Covariance matrix adaptation, Evolution strategy, Linear convergence",
author = "Tobias Glasmachers and Oswin Krause",
note = "Publisher Copyright: {\textcopyright} 2021 Massachusetts Institute of Technology.",
year = "2022",
doi = "10.1162/evco_a_00295",
language = "English",
volume = "30",
pages = "27--50",
journal = "Evolutionary Computation",
issn = "1063-6560",
publisher = "M I T Press",
number = "1",

}

RIS

TY - JOUR

T1 - Convergence Analysis of the Hessian Estimation Evolution Strategy

AU - Glasmachers, Tobias

AU - Krause, Oswin

N1 - Publisher Copyright: © 2021 Massachusetts Institute of Technology.

PY - 2022

Y1 - 2022

N2 - The class of algorithms called Hessian Estimation Evolution Strategies (HE-ESs) update the covariance matrix of their sampling distribution by directly estimating the curvature of the objective function. The approach is practically efficient, as attested by respectable performance on the BBOB testbed, even on rather irregular functions. In this article, we formally prove two strong guarantees for the (1 + 4)-HE-ES, a minimal elitist member of the family: stability of the covariance matrix update, and as a consequence, linear convergence on all convex quadratic problems at a rate that is independent of the problem instance.

AB - The class of algorithms called Hessian Estimation Evolution Strategies (HE-ESs) update the covariance matrix of their sampling distribution by directly estimating the curvature of the objective function. The approach is practically efficient, as attested by respectable performance on the BBOB testbed, even on rather irregular functions. In this article, we formally prove two strong guarantees for the (1 + 4)-HE-ES, a minimal elitist member of the family: stability of the covariance matrix update, and as a consequence, linear convergence on all convex quadratic problems at a rate that is independent of the problem instance.

KW - Covariance matrix adaptation

KW - Evolution strategy

KW - Linear convergence

U2 - 10.1162/evco_a_00295

DO - 10.1162/evco_a_00295

M3 - Journal article

C2 - 34779840

AN - SCOPUS:85125553279

VL - 30

SP - 27

EP - 50

JO - Evolutionary Computation

JF - Evolutionary Computation

SN - 1063-6560

IS - 1

ER -

ID: 307373953