Geodesic
Piecewise linear classifiers preserving high local recognition rates
Kybernetika, Tome 34 (1998) no. 4, pp. 479-484 Cet article a éte moissonné depuis la source Czech Digital Mathematics Library

Voir la notice de l'article

We propose a new method to construct piecewise linear classifiers. This method constructs hyperplanes of a piecewise linear classifier so as to keep the correct recognition rate over a threshold for a training set. The threshold is determined automatically by the MDL (Minimum Description Length) criterion so as to avoid overfitting of the classifier to the training set. The proposed method showed better results in some experiments than a previous method.
We propose a new method to construct piecewise linear classifiers. This method constructs hyperplanes of a piecewise linear classifier so as to keep the correct recognition rate over a threshold for a training set. The threshold is determined automatically by the MDL (Minimum Description Length) criterion so as to avoid overfitting of the classifier to the training set. The proposed method showed better results in some experiments than a previous method.
Classification : 68T10, 68U99
Keywords: piecewise linear classifiers; clustering; recognition rates 
@article{KYB_1998_34_4_a20,
     author = {Tenmoto, Hiroshi and Kudo, Mineichi and Shimbo, Masaru},
     title = {Piecewise linear classifiers preserving high local recognition rates},
     journal = {Kybernetika},
     pages = {479--484},
     year = {1998},
     volume = {34},
     number = {4},
     zbl = {1274.68395},
     language = {en},
     url = {http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a20/}
}
TY  - JOUR
AU  - Tenmoto, Hiroshi
AU  - Kudo, Mineichi
AU  - Shimbo, Masaru
TI  - Piecewise linear classifiers preserving high local recognition rates
JO  - Kybernetika
PY  - 1998
SP  - 479
EP  - 484
VL  - 34
IS  - 4
UR  - http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a20/
LA  - en
ID  - KYB_1998_34_4_a20
ER  - 
%0 Journal Article
%A Tenmoto, Hiroshi
%A Kudo, Mineichi
%A Shimbo, Masaru
%T Piecewise linear classifiers preserving high local recognition rates
%J Kybernetika
%D 1998
%P 479-484
%V 34
%N 4
%U http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a20/
%G en
%F KYB_1998_34_4_a20
Tenmoto, Hiroshi; Kudo, Mineichi; Shimbo, Masaru. Piecewise linear classifiers preserving high local recognition rates. Kybernetika, Tome 34 (1998) no. 4, pp. 479-484. http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a20/

[2] Forgy E. W.: Cluster analysis of multivariate data: Efficiency versus interpretability of classifications. Abstracts Biometrics 21 (1965), 3, 768

[3] al S. Hayamizu et: Generation of VCV/CVC balanced word sets for speech database. Bull. Electrotechnical Laboratory 49 (1985), 10, 803–834

[4] Mangasarian O. L.: Multisurface method of pattern separation. IEEE Trans. Inform. Theory 14 (1968), 6, 801–807 | DOI | Zbl

[5] Meisel W. S.: Computer–Oriented Approaches to Pattern Recognition. Academic Press, New York 1972 | Zbl

[6] Nilsson N. J.: Learning Machines. McGraw–Hill, New York 1965 | Zbl

[7] Park Y., Sklansky J.: Automated design of multiple–class piecewise linear classifiers. J. Classification 6 (1989), 195–222 | DOI | Zbl

[8] Rissanen J.: A universal prior for integers and estimation by minimum description length. Ann. Statist. 11 (1983), 416–431 | DOI | Zbl

[9] Sklansky J., Michelotti L.: Locally trained piecewise linear classifiers. IEEE Trans. Pattern Analysis and Machine Intelligence 2 (1980), 101–111 | DOI | Zbl

[10] Sklansky J., Wassel G. N.: Pattern Classification and Trainable Machines. Springer–Verlag, New York 1981

[11] Takiyama R.: A learning procedure for multisurface method of pattern separation. Pattern Recognition 12 (1980), 75–82 | DOI | MR | Zbl

[12] Tomek I.: Two modifications of CNN. IEEE Trans. Systems Man Cybernet. 6 (1976), 11, 769–772 | DOI | MR | Zbl

[13] Yamanishi K.: A learning criterion for stochastic rules. In: Machine Learning, to appear. An extended abstract in Proceedings of the Third Annual Workshop on Computational Learning Theory, 1990, pp. 67–81