Rule extraction from a trained neural network for image keywords extraction

H. Nishiyama, H. Kawasaki, M. Fukumi, N. Akamatsu, Yasue Mitsukura

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

Abstract

This paper presents a rule extraction method from a trained neural network (NN), which is used for keywords extraction from Images. In our approach, first, a bit map image in the RGB color space is transformed into that in the L*a*b* color space. Next, it clusters image pixels using the fuzzy c-means method and domains are extracted through a labeling process. Features, such as area of obtained domains, color information, and coordinates of the center of gravity, are then calculated, which are used as input attributes to NN. NN is then trained using such features. After NN learning, rule extraction is carried out using binarized output values in the hidden layer for each keyword. The rules extracted in this paper are If-then rules, which include logical functions. The methods of generating keywords using NN and the rules are presented and their comparative experiments are performed. Finally the validity of these methods was verified by means of computer simulations.

Original languageEnglish
Title of host publicationAnnual Conference of the North American Fuzzy Information Processing Society - NAFIPS
EditorsS.H. Dick, L. Kurgan, P. Musilek, W. Pedrycz, M. Reformat
Pages325-329
Number of pages5
Volume1
Publication statusPublished - 2004
Externally publishedYes
EventNAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society: Fuzzy Sets in the Heart of the Canadian Rockies - Banff, Alta, Canada
Duration: 2004 Jun 272004 Jun 30

Other

OtherNAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society: Fuzzy Sets in the Heart of the Canadian Rockies
CountryCanada
CityBanff, Alta
Period04/6/2704/6/30

Fingerprint

Neural networks
Color
Labeling
Gravitation
Pixels
Computer simulation
Experiments

ASJC Scopus subject areas

  • Computer Science(all)
  • Media Technology

Cite this

Nishiyama, H., Kawasaki, H., Fukumi, M., Akamatsu, N., & Mitsukura, Y. (2004). Rule extraction from a trained neural network for image keywords extraction. In S. H. Dick, L. Kurgan, P. Musilek, W. Pedrycz, & M. Reformat (Eds.), Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS (Vol. 1, pp. 325-329)

Rule extraction from a trained neural network for image keywords extraction. / Nishiyama, H.; Kawasaki, H.; Fukumi, M.; Akamatsu, N.; Mitsukura, Yasue.

Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. ed. / S.H. Dick; L. Kurgan; P. Musilek; W. Pedrycz; M. Reformat. Vol. 1 2004. p. 325-329.

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

Nishiyama, H, Kawasaki, H, Fukumi, M, Akamatsu, N & Mitsukura, Y 2004, Rule extraction from a trained neural network for image keywords extraction. in SH Dick, L Kurgan, P Musilek, W Pedrycz & M Reformat (eds), Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. vol. 1, pp. 325-329, NAFIPS 2004 - Annual Meeting of the North American Fuzzy Information Processing Society: Fuzzy Sets in the Heart of the Canadian Rockies, Banff, Alta, Canada, 04/6/27.
Nishiyama H, Kawasaki H, Fukumi M, Akamatsu N, Mitsukura Y. Rule extraction from a trained neural network for image keywords extraction. In Dick SH, Kurgan L, Musilek P, Pedrycz W, Reformat M, editors, Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. Vol. 1. 2004. p. 325-329
Nishiyama, H. ; Kawasaki, H. ; Fukumi, M. ; Akamatsu, N. ; Mitsukura, Yasue. / Rule extraction from a trained neural network for image keywords extraction. Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. editor / S.H. Dick ; L. Kurgan ; P. Musilek ; W. Pedrycz ; M. Reformat. Vol. 1 2004. pp. 325-329
@inproceedings{2a3c44393b224527830afceea20b9e13,
title = "Rule extraction from a trained neural network for image keywords extraction",
abstract = "This paper presents a rule extraction method from a trained neural network (NN), which is used for keywords extraction from Images. In our approach, first, a bit map image in the RGB color space is transformed into that in the L*a*b* color space. Next, it clusters image pixels using the fuzzy c-means method and domains are extracted through a labeling process. Features, such as area of obtained domains, color information, and coordinates of the center of gravity, are then calculated, which are used as input attributes to NN. NN is then trained using such features. After NN learning, rule extraction is carried out using binarized output values in the hidden layer for each keyword. The rules extracted in this paper are If-then rules, which include logical functions. The methods of generating keywords using NN and the rules are presented and their comparative experiments are performed. Finally the validity of these methods was verified by means of computer simulations.",
author = "H. Nishiyama and H. Kawasaki and M. Fukumi and N. Akamatsu and Yasue Mitsukura",
year = "2004",
language = "English",
volume = "1",
pages = "325--329",
editor = "S.H. Dick and L. Kurgan and P. Musilek and W. Pedrycz and M. Reformat",
booktitle = "Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS",

}

TY - GEN

T1 - Rule extraction from a trained neural network for image keywords extraction

AU - Nishiyama, H.

AU - Kawasaki, H.

AU - Fukumi, M.

AU - Akamatsu, N.

AU - Mitsukura, Yasue

PY - 2004

Y1 - 2004

N2 - This paper presents a rule extraction method from a trained neural network (NN), which is used for keywords extraction from Images. In our approach, first, a bit map image in the RGB color space is transformed into that in the L*a*b* color space. Next, it clusters image pixels using the fuzzy c-means method and domains are extracted through a labeling process. Features, such as area of obtained domains, color information, and coordinates of the center of gravity, are then calculated, which are used as input attributes to NN. NN is then trained using such features. After NN learning, rule extraction is carried out using binarized output values in the hidden layer for each keyword. The rules extracted in this paper are If-then rules, which include logical functions. The methods of generating keywords using NN and the rules are presented and their comparative experiments are performed. Finally the validity of these methods was verified by means of computer simulations.

AB - This paper presents a rule extraction method from a trained neural network (NN), which is used for keywords extraction from Images. In our approach, first, a bit map image in the RGB color space is transformed into that in the L*a*b* color space. Next, it clusters image pixels using the fuzzy c-means method and domains are extracted through a labeling process. Features, such as area of obtained domains, color information, and coordinates of the center of gravity, are then calculated, which are used as input attributes to NN. NN is then trained using such features. After NN learning, rule extraction is carried out using binarized output values in the hidden layer for each keyword. The rules extracted in this paper are If-then rules, which include logical functions. The methods of generating keywords using NN and the rules are presented and their comparative experiments are performed. Finally the validity of these methods was verified by means of computer simulations.

UR - http://www.scopus.com/inward/record.url?scp=4544248282&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=4544248282&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:4544248282

VL - 1

SP - 325

EP - 329

BT - Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS

A2 - Dick, S.H.

A2 - Kurgan, L.

A2 - Musilek, P.

A2 - Pedrycz, W.

A2 - Reformat, M.

ER -