Towards statistical convergence criteria for mutation-based evolutionary algorithms

Felipe Campelo

doi:10.1109/la-cci.2015.7435944

Towards statistical convergence criteria for mutation-based evolutionary algorithms

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

This work presents theoretical results on the development of a statistical convergence criterion for evolutionary algorithms. An analytical formula is derived for the probability of success in isotropic Gaussian mutation operators over spherical functions, and statistical criteria are proposed for evaluating, with predefined confidence levels, the convergence of (1+1) and (1+λ) Evolution Strategies. The results presented are intended as a first approach to the development of statistically based stop criteria for evolutionary optimizers, and as a contribution for the broader application of statistical modeling to the development and study of population-based algorithms.

Original language	English
Title of host publication	Proceedings of the 2015 Latin America Congress on Computational Intelligence (LA-CCI)
Publisher	IEEE
ISBN (Electronic)	978-1-4673-8417-9
ISBN (Print)	9781467384186
DOIs	https://doi.org/10.1109/la-cci.2015.7435944
Publication status	Published - 21 Mar 2016
Event	2015 Latin America Congress on Computational Intelligence (LA-CCI) - Curitiba, Brazil Duration: 13 Oct 2015 → 16 Oct 2015

Conference

Conference	2015 Latin America Congress on Computational Intelligence (LA-CCI)
Country/Territory	Brazil
City	Curitiba
Period	13/10/15 → 16/10/15

Access to Document

10.1109/la-cci.2015.7435944

Cite this

@inproceedings{c77fba4cfde54cc88e2a4876c594f9b6,

title = "Towards statistical convergence criteria for mutation-based evolutionary algorithms",

abstract = "This work presents theoretical results on the development of a statistical convergence criterion for evolutionary algorithms. An analytical formula is derived for the probability of success in isotropic Gaussian mutation operators over spherical functions, and statistical criteria are proposed for evaluating, with predefined confidence levels, the convergence of (1+1) and (1+λ) Evolution Strategies. The results presented are intended as a first approach to the development of statistically based stop criteria for evolutionary optimizers, and as a contribution for the broader application of statistical modeling to the development and study of population-based algorithms.",

author = "Felipe Campelo",

year = "2016",

month = mar,

day = "21",

doi = "10.1109/la-cci.2015.7435944",

language = "English",

isbn = "9781467384186",

booktitle = "Proceedings of the 2015 Latin America Congress on Computational Intelligence (LA-CCI)",

publisher = "IEEE",

address = "United States",

note = "2015 Latin America Congress on Computational Intelligence (LA-CCI) ; Conference date: 13-10-2015 Through 16-10-2015",

}

TY - GEN

T1 - Towards statistical convergence criteria for mutation-based evolutionary algorithms

AU - Campelo, Felipe

PY - 2016/3/21

Y1 - 2016/3/21

N2 - This work presents theoretical results on the development of a statistical convergence criterion for evolutionary algorithms. An analytical formula is derived for the probability of success in isotropic Gaussian mutation operators over spherical functions, and statistical criteria are proposed for evaluating, with predefined confidence levels, the convergence of (1+1) and (1+λ) Evolution Strategies. The results presented are intended as a first approach to the development of statistically based stop criteria for evolutionary optimizers, and as a contribution for the broader application of statistical modeling to the development and study of population-based algorithms.

AB - This work presents theoretical results on the development of a statistical convergence criterion for evolutionary algorithms. An analytical formula is derived for the probability of success in isotropic Gaussian mutation operators over spherical functions, and statistical criteria are proposed for evaluating, with predefined confidence levels, the convergence of (1+1) and (1+λ) Evolution Strategies. The results presented are intended as a first approach to the development of statistically based stop criteria for evolutionary optimizers, and as a contribution for the broader application of statistical modeling to the development and study of population-based algorithms.

UR - https://ieeexplore.ieee.org/document/7435944

U2 - 10.1109/la-cci.2015.7435944

DO - 10.1109/la-cci.2015.7435944

M3 - Conference publication

SN - 9781467384186

BT - Proceedings of the 2015 Latin America Congress on Computational Intelligence (LA-CCI)

PB - IEEE

T2 - 2015 Latin America Congress on Computational Intelligence (LA-CCI)

Y2 - 13 October 2015 through 16 October 2015

ER -

Towards statistical convergence criteria for mutation-based evolutionary algorithms

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Best Paper Award

2015 Latin America Congress on Computational Intelligence (LA-CCI)

Cite this

Towards statistical convergence criteria for mutation-based evolutionary algorithms

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Prizes

Best Paper Award

Activities

2015 Latin America Congress on Computational Intelligence (LA-CCI)

Cite this