A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance

Nicholas E. Hardison, David M. Reif, Theresa J. Fanelli, Marylyn Deriggi Ritchie, Scott M. Dudek, Alison A. Motsinger-Reif

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Grammatical Evolution Neural Networks (GENN) is a computational method designed to detect gene-gene interactions in genetic epidemiology, but has so far only been evaluated in situations with balanced numbers of cases and controls. Real data, however, rarely has such perfectly balanced classes. In the current study, we test the power of GENN to detect interactions in data with a range of class imbalance using two fitness functions (classification error and balanced error), as well as data resampling. We show that when using classification error, class imbalance greatly decreases the power of GENN. Re-sampling methods demonstrated improved power, but using balanced accuracy resulted in the highest power. Based on the results of this study, balanced error has replaced classification error in the GENN algorithm.

Original languageEnglish (US)
Title of host publicationGECCO'08
Subtitle of host publicationProceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008
Pages353-354
Number of pages2
StatePublished - Dec 15 2008
Event10th Annual Genetic and Evolutionary Computation Conference, GECCO 2008 - Atlanta, GA, United States
Duration: Jul 12 2008Jul 16 2008

Other

Other10th Annual Genetic and Evolutionary Computation Conference, GECCO 2008
CountryUnited States
CityAtlanta, GA
Period7/12/087/16/08

Fingerprint

Neural networks
Genes
Computational methods
Sampling

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Science Applications
  • Software

Cite this

Hardison, N. E., Reif, D. M., Fanelli, T. J., Ritchie, M. D., Dudek, S. M., & Motsinger-Reif, A. A. (2008). A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance. In GECCO'08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008 (pp. 353-354)
Hardison, Nicholas E. ; Reif, David M. ; Fanelli, Theresa J. ; Ritchie, Marylyn Deriggi ; Dudek, Scott M. ; Motsinger-Reif, Alison A. / A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance. GECCO'08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008. 2008. pp. 353-354
@inproceedings{14f6d3a0f0c94257aaa3b3c1c5cddba0,
title = "A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance",
abstract = "Grammatical Evolution Neural Networks (GENN) is a computational method designed to detect gene-gene interactions in genetic epidemiology, but has so far only been evaluated in situations with balanced numbers of cases and controls. Real data, however, rarely has such perfectly balanced classes. In the current study, we test the power of GENN to detect interactions in data with a range of class imbalance using two fitness functions (classification error and balanced error), as well as data resampling. We show that when using classification error, class imbalance greatly decreases the power of GENN. Re-sampling methods demonstrated improved power, but using balanced accuracy resulted in the highest power. Based on the results of this study, balanced error has replaced classification error in the GENN algorithm.",
author = "Hardison, {Nicholas E.} and Reif, {David M.} and Fanelli, {Theresa J.} and Ritchie, {Marylyn Deriggi} and Dudek, {Scott M.} and Motsinger-Reif, {Alison A.}",
year = "2008",
month = "12",
day = "15",
language = "English (US)",
isbn = "9781605581309",
pages = "353--354",
booktitle = "GECCO'08",

}

Hardison, NE, Reif, DM, Fanelli, TJ, Ritchie, MD, Dudek, SM & Motsinger-Reif, AA 2008, A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance. in GECCO'08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008. pp. 353-354, 10th Annual Genetic and Evolutionary Computation Conference, GECCO 2008, Atlanta, GA, United States, 7/12/08.

A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance. / Hardison, Nicholas E.; Reif, David M.; Fanelli, Theresa J.; Ritchie, Marylyn Deriggi; Dudek, Scott M.; Motsinger-Reif, Alison A.

GECCO'08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008. 2008. p. 353-354.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance

AU - Hardison, Nicholas E.

AU - Reif, David M.

AU - Fanelli, Theresa J.

AU - Ritchie, Marylyn Deriggi

AU - Dudek, Scott M.

AU - Motsinger-Reif, Alison A.

PY - 2008/12/15

Y1 - 2008/12/15

N2 - Grammatical Evolution Neural Networks (GENN) is a computational method designed to detect gene-gene interactions in genetic epidemiology, but has so far only been evaluated in situations with balanced numbers of cases and controls. Real data, however, rarely has such perfectly balanced classes. In the current study, we test the power of GENN to detect interactions in data with a range of class imbalance using two fitness functions (classification error and balanced error), as well as data resampling. We show that when using classification error, class imbalance greatly decreases the power of GENN. Re-sampling methods demonstrated improved power, but using balanced accuracy resulted in the highest power. Based on the results of this study, balanced error has replaced classification error in the GENN algorithm.

AB - Grammatical Evolution Neural Networks (GENN) is a computational method designed to detect gene-gene interactions in genetic epidemiology, but has so far only been evaluated in situations with balanced numbers of cases and controls. Real data, however, rarely has such perfectly balanced classes. In the current study, we test the power of GENN to detect interactions in data with a range of class imbalance using two fitness functions (classification error and balanced error), as well as data resampling. We show that when using classification error, class imbalance greatly decreases the power of GENN. Re-sampling methods demonstrated improved power, but using balanced accuracy resulted in the highest power. Based on the results of this study, balanced error has replaced classification error in the GENN algorithm.

UR - http://www.scopus.com/inward/record.url?scp=57349087730&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=57349087730&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781605581309

SP - 353

EP - 354

BT - GECCO'08

ER -

Hardison NE, Reif DM, Fanelli TJ, Ritchie MD, Dudek SM, Motsinger-Reif AA. A balanced accuracy fitness function leads to robust analysis using grammatical evolution neural networks in the case of class imbalance. In GECCO'08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation 2008. 2008. p. 353-354