A new virtual-sample-generating method based on the heuristics algorithm

Der-Chiang Li, I. Hsiang Wen, Chih Chieh Chang

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

1 Citation (Scopus)

Abstract

While back-propagation neural networks (BPNN) are effective learning tools for building non-linear models, they are often unstable when using small-data-sets. Therefore, in order to solve this problem, we construct artificial samples, called virtual samples, to improve the learning robustness. This research develops a novel method of virtual sample generation (VSG), named genetic algorithm-based virtual sample generation (GABVSG), which considers the integrated effects and constraints of data attributes. We first determine the acceptable range by using MTD functions, and construct the feasibility-based programming (FBP) model with BPNN. A genetic algorithm (GA) is then applied to accelerate the generation of feasible virtual samples. Finally, we use two real cases to verify the performance of the proposed method by comparing the results with those of two forecasting models, BPNN and support vector machine for regression (SVR). The experimental results indicate that the performance of the GABVSG method is superior to that of using original training data without artificial samples. Consequently, the proposed method can improve learning performance significantly when working with small samples.

Original languageEnglish
Title of host publicationProceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013
Pages469-472
Number of pages4
DOIs
Publication statusPublished - 2013
Event2013 24th IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013 - Macau, China
Duration: 2013 Nov 152013 Nov 17

Other

Other2013 24th IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013
CountryChina
CityMacau
Period13-11-1513-11-17

Fingerprint

Heuristic algorithms
Backpropagation
Genetic algorithms
Neural networks
Support vector machines

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Science Applications
  • Hardware and Architecture
  • Control and Systems Engineering

Cite this

Li, D-C., Wen, I. H., & Chang, C. C. (2013). A new virtual-sample-generating method based on the heuristics algorithm. In Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013 (pp. 469-472). [6714829] https://doi.org/10.1109/GSIS.2013.6714829
Li, Der-Chiang ; Wen, I. Hsiang ; Chang, Chih Chieh. / A new virtual-sample-generating method based on the heuristics algorithm. Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013. 2013. pp. 469-472
@inproceedings{238ff7c0a6d34d50b2a78b43fd1128a7,
title = "A new virtual-sample-generating method based on the heuristics algorithm",
abstract = "While back-propagation neural networks (BPNN) are effective learning tools for building non-linear models, they are often unstable when using small-data-sets. Therefore, in order to solve this problem, we construct artificial samples, called virtual samples, to improve the learning robustness. This research develops a novel method of virtual sample generation (VSG), named genetic algorithm-based virtual sample generation (GABVSG), which considers the integrated effects and constraints of data attributes. We first determine the acceptable range by using MTD functions, and construct the feasibility-based programming (FBP) model with BPNN. A genetic algorithm (GA) is then applied to accelerate the generation of feasible virtual samples. Finally, we use two real cases to verify the performance of the proposed method by comparing the results with those of two forecasting models, BPNN and support vector machine for regression (SVR). The experimental results indicate that the performance of the GABVSG method is superior to that of using original training data without artificial samples. Consequently, the proposed method can improve learning performance significantly when working with small samples.",
author = "Der-Chiang Li and Wen, {I. Hsiang} and Chang, {Chih Chieh}",
year = "2013",
doi = "10.1109/GSIS.2013.6714829",
language = "English",
isbn = "9781467352628",
pages = "469--472",
booktitle = "Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013",

}

Li, D-C, Wen, IH & Chang, CC 2013, A new virtual-sample-generating method based on the heuristics algorithm. in Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013., 6714829, pp. 469-472, 2013 24th IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013, Macau, China, 13-11-15. https://doi.org/10.1109/GSIS.2013.6714829

A new virtual-sample-generating method based on the heuristics algorithm. / Li, Der-Chiang; Wen, I. Hsiang; Chang, Chih Chieh.

Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013. 2013. p. 469-472 6714829.

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

TY - GEN

T1 - A new virtual-sample-generating method based on the heuristics algorithm

AU - Li, Der-Chiang

AU - Wen, I. Hsiang

AU - Chang, Chih Chieh

PY - 2013

Y1 - 2013

N2 - While back-propagation neural networks (BPNN) are effective learning tools for building non-linear models, they are often unstable when using small-data-sets. Therefore, in order to solve this problem, we construct artificial samples, called virtual samples, to improve the learning robustness. This research develops a novel method of virtual sample generation (VSG), named genetic algorithm-based virtual sample generation (GABVSG), which considers the integrated effects and constraints of data attributes. We first determine the acceptable range by using MTD functions, and construct the feasibility-based programming (FBP) model with BPNN. A genetic algorithm (GA) is then applied to accelerate the generation of feasible virtual samples. Finally, we use two real cases to verify the performance of the proposed method by comparing the results with those of two forecasting models, BPNN and support vector machine for regression (SVR). The experimental results indicate that the performance of the GABVSG method is superior to that of using original training data without artificial samples. Consequently, the proposed method can improve learning performance significantly when working with small samples.

AB - While back-propagation neural networks (BPNN) are effective learning tools for building non-linear models, they are often unstable when using small-data-sets. Therefore, in order to solve this problem, we construct artificial samples, called virtual samples, to improve the learning robustness. This research develops a novel method of virtual sample generation (VSG), named genetic algorithm-based virtual sample generation (GABVSG), which considers the integrated effects and constraints of data attributes. We first determine the acceptable range by using MTD functions, and construct the feasibility-based programming (FBP) model with BPNN. A genetic algorithm (GA) is then applied to accelerate the generation of feasible virtual samples. Finally, we use two real cases to verify the performance of the proposed method by comparing the results with those of two forecasting models, BPNN and support vector machine for regression (SVR). The experimental results indicate that the performance of the GABVSG method is superior to that of using original training data without artificial samples. Consequently, the proposed method can improve learning performance significantly when working with small samples.

UR - http://www.scopus.com/inward/record.url?scp=84897885509&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84897885509&partnerID=8YFLogxK

U2 - 10.1109/GSIS.2013.6714829

DO - 10.1109/GSIS.2013.6714829

M3 - Conference contribution

AN - SCOPUS:84897885509

SN - 9781467352628

SP - 469

EP - 472

BT - Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013

ER -

Li D-C, Wen IH, Chang CC. A new virtual-sample-generating method based on the heuristics algorithm. In Proceedings of 2013 IEEE International Conference on Grey Systems and Intelligent Services, GSIS 2013. 2013. p. 469-472. 6714829 https://doi.org/10.1109/GSIS.2013.6714829