Mirror image learning for handwritten numeral recognition

Meng Shi; Tetsushi Wakabayashi; Wataru Ohyama; Fumitaka Kimura

doi:10.1007/3-540-44596-x_20

Mirror image learning for handwritten numeral recognition

Meng Shi, Tetsushi Wakabayashi, Wataru Ohyama, Fumitaka Kimura

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

3 Citations (Scopus)

Abstract

This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair of confusing classes and utilizes it as a learning pattern of the other class. Statistical pattern recognition techniques generally assume that the density function and the parameters of each class are only dependent on the sample of the class. The mirror image learning algorithm enlarges the learning sample of each class by mirror image patterns of other classes and enables us to achieve higher recognition accuracy with small learning sample. Recognition accuracies of the minimum distance classifier and the projection distance method were improved from 93.17% to 98.38% and from 99.11% to 99.37% respectively in the recognition test for handwritten numeral database IPTP CD-ROM1 [1].

Original language	English
Title of host publication	Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings
Editors	Petra Perner
Publisher	Springer Verlag
Pages	239-248
Number of pages	10
ISBN (Print)	3540423591, 9783540423591
DOIs	https://doi.org/10.1007/3-540-44596-x_20
Publication status	Published - 2001
Externally published	Yes
Event	2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001 - Leipzig, Germany Duration: Jul 25 2001 → Jul 27 2001

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	2123 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001
Country/Territory	Germany
City	Leipzig
Period	7/25/01 → 7/27/01

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/3-540-44596-x_20

Cite this

Shi, M., Wakabayashi, T., Ohyama, W., & Kimura, F. (2001). Mirror image learning for handwritten numeral recognition. In P. Perner (Ed.), Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings (pp. 239-248). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2123 LNAI). Springer Verlag. https://doi.org/10.1007/3-540-44596-x_20

Mirror image learning for handwritten numeral recognition. / Shi, Meng; Wakabayashi, Tetsushi; Ohyama, Wataru et al.
Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings. ed. / Petra Perner. Springer Verlag, 2001. p. 239-248 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2123 LNAI).

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

Shi, M, Wakabayashi, T, Ohyama, W & Kimura, F 2001, Mirror image learning for handwritten numeral recognition. in P Perner (ed.), Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 2123 LNAI, Springer Verlag, pp. 239-248, 2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001, Leipzig, Germany, 7/25/01. https://doi.org/10.1007/3-540-44596-x_20

Shi M, Wakabayashi T, Ohyama W, Kimura F. Mirror image learning for handwritten numeral recognition. In Perner P, editor, Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings. Springer Verlag. 2001. p. 239-248. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/3-540-44596-x_20

Shi, Meng ; Wakabayashi, Tetsushi ; Ohyama, Wataru et al. / Mirror image learning for handwritten numeral recognition. Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings. editor / Petra Perner. Springer Verlag, 2001. pp. 239-248 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{22d6f05f89c643e9a8a93d46a18cef5a,

title = "Mirror image learning for handwritten numeral recognition",

abstract = "This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair of confusing classes and utilizes it as a learning pattern of the other class. Statistical pattern recognition techniques generally assume that the density function and the parameters of each class are only dependent on the sample of the class. The mirror image learning algorithm enlarges the learning sample of each class by mirror image patterns of other classes and enables us to achieve higher recognition accuracy with small learning sample. Recognition accuracies of the minimum distance classifier and the projection distance method were improved from 93.17% to 98.38% and from 99.11% to 99.37% respectively in the recognition test for handwritten numeral database IPTP CD-ROM1 [1].",

author = "Meng Shi and Tetsushi Wakabayashi and Wataru Ohyama and Fumitaka Kimura",

note = "Copyright: Copyright 2020 Elsevier B.V., All rights reserved.; 2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001 ; Conference date: 25-07-2001 Through 27-07-2001",

year = "2001",

doi = "10.1007/3-540-44596-x_20",

language = "English",

isbn = "3540423591",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "239--248",

editor = "Petra Perner",

booktitle = "Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings",

address = "Germany",

}

TY - GEN

T1 - Mirror image learning for handwritten numeral recognition

AU - Shi, Meng

AU - Wakabayashi, Tetsushi

AU - Ohyama, Wataru

AU - Kimura, Fumitaka

PY - 2001

Y1 - 2001

N2 - This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair of confusing classes and utilizes it as a learning pattern of the other class. Statistical pattern recognition techniques generally assume that the density function and the parameters of each class are only dependent on the sample of the class. The mirror image learning algorithm enlarges the learning sample of each class by mirror image patterns of other classes and enables us to achieve higher recognition accuracy with small learning sample. Recognition accuracies of the minimum distance classifier and the projection distance method were improved from 93.17% to 98.38% and from 99.11% to 99.37% respectively in the recognition test for handwritten numeral database IPTP CD-ROM1 [1].

AB - This paper proposes a new corrective learning algorithm and evaluates the performance by handwritten numeral recognition test. The algorithm generates a mirror image of a pattern which belongs to one class of a pair of confusing classes and utilizes it as a learning pattern of the other class. Statistical pattern recognition techniques generally assume that the density function and the parameters of each class are only dependent on the sample of the class. The mirror image learning algorithm enlarges the learning sample of each class by mirror image patterns of other classes and enables us to achieve higher recognition accuracy with small learning sample. Recognition accuracies of the minimum distance classifier and the projection distance method were improved from 93.17% to 98.38% and from 99.11% to 99.37% respectively in the recognition test for handwritten numeral database IPTP CD-ROM1 [1].

UR - http://www.scopus.com/inward/record.url?scp=84899461862&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84899461862&partnerID=8YFLogxK

U2 - 10.1007/3-540-44596-x_20

DO - 10.1007/3-540-44596-x_20

M3 - Conference contribution

AN - SCOPUS:84899461862

SN - 3540423591

SN - 9783540423591

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 239

EP - 248

BT - Machine Learning and Data Mining in Pattern Recognition - Second International Workshop, MLDM 2001, Proceedings

A2 - Perner, Petra

PB - Springer Verlag

T2 - 2nd International Workshop on Machine Learning and Data Mining in Pattern Recognition, MLDM 2001

Y2 - 25 July 2001 through 27 July 2001

ER -

Mirror image learning for handwritten numeral recognition

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this