A new recurrent neural network architecture for pattern recognition

Hee Heon Song, Sun Mee Kang, Seong Whan Lee

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

3 Citations (Scopus)

Abstract

In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent neural network differs from Jordan's and Elman's recurrent neural networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm in the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.

Original languageEnglish
Title of host publicationProceedings - International Conference on Pattern Recognition
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages718-722
Number of pages5
Volume4
ISBN (Print)081867282X, 9780818672828
DOIs
Publication statusPublished - 1996 Jan 1
Event13th International Conference on Pattern Recognition, ICPR 1996 - Vienna, Austria
Duration: 1996 Aug 251996 Aug 29

Other

Other13th International Conference on Pattern Recognition, ICPR 1996
CountryAustria
CityVienna
Period96/8/2596/8/29

Fingerprint

Recurrent neural networks
Network architecture
Pattern recognition
Feedforward neural networks
Multilayer neural networks
Learning algorithms
Experiments

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Song, H. H., Kang, S. M., & Lee, S. W. (1996). A new recurrent neural network architecture for pattern recognition. In Proceedings - International Conference on Pattern Recognition (Vol. 4, pp. 718-722). [547658] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPR.1996.547658

A new recurrent neural network architecture for pattern recognition. / Song, Hee Heon; Kang, Sun Mee; Lee, Seong Whan.

Proceedings - International Conference on Pattern Recognition. Vol. 4 Institute of Electrical and Electronics Engineers Inc., 1996. p. 718-722 547658.

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

Song, HH, Kang, SM & Lee, SW 1996, A new recurrent neural network architecture for pattern recognition. in Proceedings - International Conference on Pattern Recognition. vol. 4, 547658, Institute of Electrical and Electronics Engineers Inc., pp. 718-722, 13th International Conference on Pattern Recognition, ICPR 1996, Vienna, Austria, 96/8/25. https://doi.org/10.1109/ICPR.1996.547658
Song HH, Kang SM, Lee SW. A new recurrent neural network architecture for pattern recognition. In Proceedings - International Conference on Pattern Recognition. Vol. 4. Institute of Electrical and Electronics Engineers Inc. 1996. p. 718-722. 547658 https://doi.org/10.1109/ICPR.1996.547658
Song, Hee Heon ; Kang, Sun Mee ; Lee, Seong Whan. / A new recurrent neural network architecture for pattern recognition. Proceedings - International Conference on Pattern Recognition. Vol. 4 Institute of Electrical and Electronics Engineers Inc., 1996. pp. 718-722
@inproceedings{c0a049a108144fe8a33e79fc07974803,
title = "A new recurrent neural network architecture for pattern recognition",
abstract = "In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent neural network differs from Jordan's and Elman's recurrent neural networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm in the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.",
author = "Song, {Hee Heon} and Kang, {Sun Mee} and Lee, {Seong Whan}",
year = "1996",
month = "1",
day = "1",
doi = "10.1109/ICPR.1996.547658",
language = "English",
isbn = "081867282X",
volume = "4",
pages = "718--722",
booktitle = "Proceedings - International Conference on Pattern Recognition",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - A new recurrent neural network architecture for pattern recognition

AU - Song, Hee Heon

AU - Kang, Sun Mee

AU - Lee, Seong Whan

PY - 1996/1/1

Y1 - 1996/1/1

N2 - In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent neural network differs from Jordan's and Elman's recurrent neural networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm in the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.

AB - In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent neural network differs from Jordan's and Elman's recurrent neural networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm in the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.

UR - http://www.scopus.com/inward/record.url?scp=0009054509&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0009054509&partnerID=8YFLogxK

U2 - 10.1109/ICPR.1996.547658

DO - 10.1109/ICPR.1996.547658

M3 - Conference contribution

AN - SCOPUS:0009054509

SN - 081867282X

SN - 9780818672828

VL - 4

SP - 718

EP - 722

BT - Proceedings - International Conference on Pattern Recognition

PB - Institute of Electrical and Electronics Engineers Inc.

ER -