Topology preserving SOM with transductive confidence machine

Bin Tong, Zhiguang Qin, Einoshin Suzuki

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

3 Citations (Scopus)

Abstract

We propose a novel topology preserving self-organized map (SOM) classifier with transductive confidence machine (TPSOM-TCM). Typically, SOM acts as a dimension reduction tool for mapping training samples from a high-dimensional input space onto a neuron grid. However, current SOM-based classifiers can not provide degrees of classification reliability for new unlabeled samples so that they are difficult to be used in risk-sensitive applications where incorrect predictions may result in serious consequences. Our method extends a typical SOM classifier to allow it to supply such reliability degrees. To achieve this objective, we define a nonconformity measurement with which a randomness test can predict how nonconforming a new unlabeled sample is with respect to the training samples. In addition, we notice that the definition of nonconformity measurement is more dependent on the quality of topology preservation than that of quantization error reduction. We thus incorporate the grey relation coefficient (GRC) into the calculation of neighborhood radii to improve the topology preservation without increasing the quantization error. Our method is able to improve the time efficiency of a previous method kNN-TCM, when the number of samples is large. Extensive experiments on both the UCI and KDDCUP 99 data sets show the effectiveness of our method.

Original languageEnglish
Title of host publicationDiscovery Science - 13th International Conference, DS 2010, Proceedings
Pages27-41
Number of pages15
DOIs
Publication statusPublished - Dec 20 2010
Event13th International Conference on Discovery Science, DS 2010 - Canberra, ACT, Australia
Duration: Oct 6 2010Oct 8 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6332 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th International Conference on Discovery Science, DS 2010
CountryAustralia
CityCanberra, ACT
Period10/6/1010/8/10

Fingerprint

Confidence
Topology
Classifiers
Topology Preservation
Classifier
Training Samples
Quantization
Error Reduction
Neurons
Dimension Reduction
Randomness
Neuron
High-dimensional
Radius
Grid
Predict
Dependent
Prediction
Coefficient
Experiments

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Tong, B., Qin, Z., & Suzuki, E. (2010). Topology preserving SOM with transductive confidence machine. In Discovery Science - 13th International Conference, DS 2010, Proceedings (pp. 27-41). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6332 LNAI). https://doi.org/10.1007/978-3-642-16184-1_3

Topology preserving SOM with transductive confidence machine. / Tong, Bin; Qin, Zhiguang; Suzuki, Einoshin.

Discovery Science - 13th International Conference, DS 2010, Proceedings. 2010. p. 27-41 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6332 LNAI).

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

Tong, B, Qin, Z & Suzuki, E 2010, Topology preserving SOM with transductive confidence machine. in Discovery Science - 13th International Conference, DS 2010, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6332 LNAI, pp. 27-41, 13th International Conference on Discovery Science, DS 2010, Canberra, ACT, Australia, 10/6/10. https://doi.org/10.1007/978-3-642-16184-1_3
Tong B, Qin Z, Suzuki E. Topology preserving SOM with transductive confidence machine. In Discovery Science - 13th International Conference, DS 2010, Proceedings. 2010. p. 27-41. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-16184-1_3
Tong, Bin ; Qin, Zhiguang ; Suzuki, Einoshin. / Topology preserving SOM with transductive confidence machine. Discovery Science - 13th International Conference, DS 2010, Proceedings. 2010. pp. 27-41 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{e73e015cc02245639b8383da0f210a25,
title = "Topology preserving SOM with transductive confidence machine",
abstract = "We propose a novel topology preserving self-organized map (SOM) classifier with transductive confidence machine (TPSOM-TCM). Typically, SOM acts as a dimension reduction tool for mapping training samples from a high-dimensional input space onto a neuron grid. However, current SOM-based classifiers can not provide degrees of classification reliability for new unlabeled samples so that they are difficult to be used in risk-sensitive applications where incorrect predictions may result in serious consequences. Our method extends a typical SOM classifier to allow it to supply such reliability degrees. To achieve this objective, we define a nonconformity measurement with which a randomness test can predict how nonconforming a new unlabeled sample is with respect to the training samples. In addition, we notice that the definition of nonconformity measurement is more dependent on the quality of topology preservation than that of quantization error reduction. We thus incorporate the grey relation coefficient (GRC) into the calculation of neighborhood radii to improve the topology preservation without increasing the quantization error. Our method is able to improve the time efficiency of a previous method kNN-TCM, when the number of samples is large. Extensive experiments on both the UCI and KDDCUP 99 data sets show the effectiveness of our method.",
author = "Bin Tong and Zhiguang Qin and Einoshin Suzuki",
year = "2010",
month = "12",
day = "20",
doi = "10.1007/978-3-642-16184-1_3",
language = "English",
isbn = "3642161839",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "27--41",
booktitle = "Discovery Science - 13th International Conference, DS 2010, Proceedings",

}

TY - GEN

T1 - Topology preserving SOM with transductive confidence machine

AU - Tong, Bin

AU - Qin, Zhiguang

AU - Suzuki, Einoshin

PY - 2010/12/20

Y1 - 2010/12/20

N2 - We propose a novel topology preserving self-organized map (SOM) classifier with transductive confidence machine (TPSOM-TCM). Typically, SOM acts as a dimension reduction tool for mapping training samples from a high-dimensional input space onto a neuron grid. However, current SOM-based classifiers can not provide degrees of classification reliability for new unlabeled samples so that they are difficult to be used in risk-sensitive applications where incorrect predictions may result in serious consequences. Our method extends a typical SOM classifier to allow it to supply such reliability degrees. To achieve this objective, we define a nonconformity measurement with which a randomness test can predict how nonconforming a new unlabeled sample is with respect to the training samples. In addition, we notice that the definition of nonconformity measurement is more dependent on the quality of topology preservation than that of quantization error reduction. We thus incorporate the grey relation coefficient (GRC) into the calculation of neighborhood radii to improve the topology preservation without increasing the quantization error. Our method is able to improve the time efficiency of a previous method kNN-TCM, when the number of samples is large. Extensive experiments on both the UCI and KDDCUP 99 data sets show the effectiveness of our method.

AB - We propose a novel topology preserving self-organized map (SOM) classifier with transductive confidence machine (TPSOM-TCM). Typically, SOM acts as a dimension reduction tool for mapping training samples from a high-dimensional input space onto a neuron grid. However, current SOM-based classifiers can not provide degrees of classification reliability for new unlabeled samples so that they are difficult to be used in risk-sensitive applications where incorrect predictions may result in serious consequences. Our method extends a typical SOM classifier to allow it to supply such reliability degrees. To achieve this objective, we define a nonconformity measurement with which a randomness test can predict how nonconforming a new unlabeled sample is with respect to the training samples. In addition, we notice that the definition of nonconformity measurement is more dependent on the quality of topology preservation than that of quantization error reduction. We thus incorporate the grey relation coefficient (GRC) into the calculation of neighborhood radii to improve the topology preservation without increasing the quantization error. Our method is able to improve the time efficiency of a previous method kNN-TCM, when the number of samples is large. Extensive experiments on both the UCI and KDDCUP 99 data sets show the effectiveness of our method.

UR - http://www.scopus.com/inward/record.url?scp=78650133003&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650133003&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-16184-1_3

DO - 10.1007/978-3-642-16184-1_3

M3 - Conference contribution

AN - SCOPUS:78650133003

SN - 3642161839

SN - 9783642161834

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

SP - 27

EP - 41

BT - Discovery Science - 13th International Conference, DS 2010, Proceedings

ER -