Constructive Induction and genetic algorithms for learning concepts with complex interaction

Leila Shila Shafti; Eduardo Pérez Pérez

doi:10.1145/1068009.1068317

Constructive Induction and genetic algorithms for learning concepts with complex interaction

Leila Shila Shafti^*, Eduardo Pérez Pérez

^*Corresponding author for this work

Computer And Information Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

15 Citations (Scopus)

Abstract

Constructive Induction is the process of transforming the original representation of hard concepts with complex interaction into a representation that highlights regularities. Most Constructive Induction methods apply a greedy strategy to find interacting attributes and then construct functions over them. This approach fails when complex interaction exists among attributes and the search space has high variation. In this paper, we illustrate the importance of applying Genetic Algorithms as a global search strategy for these methods and present MFE2/GA¹, while comparing it with other GA-based Constructive Induction methods. We empirically analyze our Genetic Algorithm's operators and compare MFE2/GA with greedy-based methods. We also performed experiments to evaluate the presented method when concept has attributes participating in more than one complex interaction. In experiments that are conducted, MFE2/GA successfully finds interacting attributes and constructs functions to represent interactions. Results show the advantage of using Genetic Algorithms for Constructive Induction when compared with greedy-based methods.

Original language	English
Title of host publication	GECCO 2005 - Genetic and Evolutionary Computation Conference
Editors	H.G. Beyer, U.M. O'Reilly, D. Arnold, W. Banzhaf, C. Blum, E.W. Bonabeau, E. Cantu-Paz, D. Dasgupta, K. Deb, al et al
Publisher	Association for Computing Machinery (ACM)
Pages	1811-1818
Number of pages	8
ISBN (Electronic)	978-1-59593-010-1
ISBN (Print)	1595930108
DOIs	https://doi.org/10.1145/1068009.1068317
Publication status	Published - 25 Jun 2005
Event	GECCO 2005 - Genetic and Evolutionary Computation Conference - Washington, D.C., United States Duration: 25 Jun 2005 → 29 Jun 2005

Conference

Conference	GECCO 2005 - Genetic and Evolutionary Computation Conference
Country/Territory	United States
City	Washington, D.C.
Period	25/06/05 → 29/06/05

Keywords

Attribute interaction
Constructive induction
Feature construction
Feature selection
Genetic algorithms
Shared attributes

Access to Document

10.1145/1068009.1068317

https://repositorio.uam.es/bitstream/handle/10486/664810/constructive_shafti_GECCO_2005_ps.pdf;jsessionid=ECB8C791F69A52D02118F172AD1FBEBF?sequence=1

Cite this

Shafti, L. S., & Pérez, E. P. (2005). Constructive Induction and genetic algorithms for learning concepts with complex interaction. In H. G. Beyer, U. M. O'Reilly, D. Arnold, W. Banzhaf, C. Blum, E. W. Bonabeau, E. Cantu-Paz, D. Dasgupta, K. Deb, & A. et al (Eds.), GECCO 2005 - Genetic and Evolutionary Computation Conference (pp. 1811-1818). Association for Computing Machinery (ACM). https://doi.org/10.1145/1068009.1068317

Shafti, Leila Shila ; Pérez, Eduardo Pérez. / Constructive Induction and genetic algorithms for learning concepts with complex interaction. GECCO 2005 - Genetic and Evolutionary Computation Conference. editor / H.G. Beyer ; U.M. O'Reilly ; D. Arnold ; W. Banzhaf ; C. Blum ; E.W. Bonabeau ; E. Cantu-Paz ; D. Dasgupta ; K. Deb ; al et al. Association for Computing Machinery (ACM), 2005. pp. 1811-1818

@inproceedings{ae9304a6b2e04f9ab917c4b010a05785,

title = "Constructive Induction and genetic algorithms for learning concepts with complex interaction",

abstract = "Constructive Induction is the process of transforming the original representation of hard concepts with complex interaction into a representation that highlights regularities. Most Constructive Induction methods apply a greedy strategy to find interacting attributes and then construct functions over them. This approach fails when complex interaction exists among attributes and the search space has high variation. In this paper, we illustrate the importance of applying Genetic Algorithms as a global search strategy for these methods and present MFE2/GA1, while comparing it with other GA-based Constructive Induction methods. We empirically analyze our Genetic Algorithm's operators and compare MFE2/GA with greedy-based methods. We also performed experiments to evaluate the presented method when concept has attributes participating in more than one complex interaction. In experiments that are conducted, MFE2/GA successfully finds interacting attributes and constructs functions to represent interactions. Results show the advantage of using Genetic Algorithms for Constructive Induction when compared with greedy-based methods.",

keywords = "Attribute interaction, Constructive induction, Feature construction, Feature selection, Genetic algorithms, Shared attributes",

author = "Shafti, \{Leila Shila\} and P{\'e}rez, \{Eduardo P{\'e}rez\}",

year = "2005",

month = jun,

day = "25",

doi = "10.1145/1068009.1068317",

language = "English",

isbn = "1595930108",

pages = "1811--1818",

editor = "H.G. Beyer and U.M. O'Reilly and D. Arnold and W. Banzhaf and C. Blum and E.W. Bonabeau and E. Cantu-Paz and D. Dasgupta and K. Deb and \{et al\}, al",

booktitle = "GECCO 2005 - Genetic and Evolutionary Computation Conference",

publisher = "Association for Computing Machinery (ACM)",

address = "United States",

note = "GECCO 2005 - Genetic and Evolutionary Computation Conference ; Conference date: 25-06-2005 Through 29-06-2005",

}

Shafti, LS & Pérez, EP 2005, Constructive Induction and genetic algorithms for learning concepts with complex interaction. in HG Beyer, UM O'Reilly, D Arnold, W Banzhaf, C Blum, EW Bonabeau, E Cantu-Paz, D Dasgupta, K Deb & A et al (eds), GECCO 2005 - Genetic and Evolutionary Computation Conference. Association for Computing Machinery (ACM), pp. 1811-1818, GECCO 2005 - Genetic and Evolutionary Computation Conference, Washington, D.C., United States, 25/06/05. https://doi.org/10.1145/1068009.1068317

Constructive Induction and genetic algorithms for learning concepts with complex interaction. / Shafti, Leila Shila; Pérez, Eduardo Pérez.
GECCO 2005 - Genetic and Evolutionary Computation Conference. ed. / H.G. Beyer; U.M. O'Reilly; D. Arnold; W. Banzhaf; C. Blum; E.W. Bonabeau; E. Cantu-Paz; D. Dasgupta; K. Deb; al et al. Association for Computing Machinery (ACM), 2005. p. 1811-1818.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

TY - GEN

T1 - Constructive Induction and genetic algorithms for learning concepts with complex interaction

AU - Shafti, Leila Shila

AU - Pérez, Eduardo Pérez

PY - 2005/6/25

Y1 - 2005/6/25

N2 - Constructive Induction is the process of transforming the original representation of hard concepts with complex interaction into a representation that highlights regularities. Most Constructive Induction methods apply a greedy strategy to find interacting attributes and then construct functions over them. This approach fails when complex interaction exists among attributes and the search space has high variation. In this paper, we illustrate the importance of applying Genetic Algorithms as a global search strategy for these methods and present MFE2/GA1, while comparing it with other GA-based Constructive Induction methods. We empirically analyze our Genetic Algorithm's operators and compare MFE2/GA with greedy-based methods. We also performed experiments to evaluate the presented method when concept has attributes participating in more than one complex interaction. In experiments that are conducted, MFE2/GA successfully finds interacting attributes and constructs functions to represent interactions. Results show the advantage of using Genetic Algorithms for Constructive Induction when compared with greedy-based methods.

AB - Constructive Induction is the process of transforming the original representation of hard concepts with complex interaction into a representation that highlights regularities. Most Constructive Induction methods apply a greedy strategy to find interacting attributes and then construct functions over them. This approach fails when complex interaction exists among attributes and the search space has high variation. In this paper, we illustrate the importance of applying Genetic Algorithms as a global search strategy for these methods and present MFE2/GA1, while comparing it with other GA-based Constructive Induction methods. We empirically analyze our Genetic Algorithm's operators and compare MFE2/GA with greedy-based methods. We also performed experiments to evaluate the presented method when concept has attributes participating in more than one complex interaction. In experiments that are conducted, MFE2/GA successfully finds interacting attributes and constructs functions to represent interactions. Results show the advantage of using Genetic Algorithms for Constructive Induction when compared with greedy-based methods.

KW - Attribute interaction

KW - Constructive induction

KW - Feature construction

KW - Feature selection

KW - Genetic algorithms

KW - Shared attributes

UR - https://www.scopus.com/pages/publications/32444442471

U2 - 10.1145/1068009.1068317

DO - 10.1145/1068009.1068317

M3 - Conference contribution book

AN - SCOPUS:32444442471

SN - 1595930108

SP - 1811

EP - 1818

BT - GECCO 2005 - Genetic and Evolutionary Computation Conference

A2 - Beyer, H.G.

A2 - O'Reilly, U.M.

A2 - Arnold, D.

A2 - Banzhaf, W.

A2 - Blum, C.

A2 - Bonabeau, E.W.

A2 - Cantu-Paz, E.

A2 - Dasgupta, D.

A2 - Deb, K.

A2 - et al, al

PB - Association for Computing Machinery (ACM)

T2 - GECCO 2005 - Genetic and Evolutionary Computation Conference

Y2 - 25 June 2005 through 29 June 2005

ER -

Shafti LS, Pérez EP. Constructive Induction and genetic algorithms for learning concepts with complex interaction. In Beyer HG, O'Reilly UM, Arnold D, Banzhaf W, Blum C, Bonabeau EW, Cantu-Paz E, Dasgupta D, Deb K, et al A, editors, GECCO 2005 - Genetic and Evolutionary Computation Conference. Association for Computing Machinery (ACM). 2005. p. 1811-1818 doi: 10.1145/1068009.1068317

Constructive Induction and genetic algorithms for learning concepts with complex interaction

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this