An Empirical Study of Progressive Insular Cooperative GP

Research output: Contribution to journalArticlepeer-review

Abstract

Genetic programming (GP) is a general purpose artificial intelligence method, that breeds populations of computer programs to solve a given problem, mimicking the principles of Darwinian evolution. Among the several different uses, it can be employed for supervised machine learning, interpreting the evolving programs as predictive models. With the objective of improving GP for multi-class classification, in this paper we model a feature of biological evolution: the structuring of populations into sub-populations, or demes. In particular, we present the progressively insular cooperative GP (PIC GP), in which classification is performed by applying two stages, in two different co-evolving sub-populations: a population, called population of specialists, aimed at optimizing the learning for the different classes, and a population, called population of teams, in which specialists are joined and the evolution allows us to obtain the final predictive model. By means of three simple parameters, PIC GP can tune the amount of cooperation between specialists of different classes. Preliminary experiments indicate that PIC GP achieves the best performance when the evolution begins with a high level of cooperation between specialists of different classes, and then this type of cooperation is progressively decreased, until only specialists of the same class can cooperate between each other. In this paper, we compare PIC GP with some state-of-the-art classification algorithms on a rich set of test applications. The obtained results show that PIC GP is highly competitive with the best algorithms, outperforming the majority of its competitors on the studied problems.
Original languageEnglish
Article number119
Pages (from-to)1-16
Number of pages16
JournalSN Computer Science
Volume3
Issue number2
DOIs
Publication statusPublished - 1 Mar 2022

Keywords

  • Multiclass classifcation
  • Genetic programming
  • Cooperative evolution

Fingerprint

Dive into the research topics of 'An Empirical Study of Progressive Insular Cooperative GP'. Together they form a unique fingerprint.

Cite this