Reducing hubness for kernel regression

Kazuo Hara; Ikumi Suzuki; Kei Kobayashi; Kenji Fukumizu; Miloš Radovanović

doi:10.1007/978-3-319-25087-8_33

Reducing hubness for kernel regression

Kazuo Hara, Ikumi Suzuki, Kei Kobayashi, Kenji Fukumizu, Miloš Radovanović

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

Abstract

In this paper, we point out that hubness—some samples in a high-dimensional dataset emerge as hubs that are similar to many other samples—influences the performance of kernel regression. Because the dimension of feature spaces induced by kernels is usually very high, hubness occurs, giving rise to the problem of multicollinearity, which is known as a cause of instability of regression results. We propose hubnessreduced kernels for kernel regression as an extension of a previous approach for kNN classification that reduces spatial centrality to eliminate hubness.

Original language	English
Title of host publication	Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings
Editors	Richard Connor, Giuseppe Amato, Fabrizio Falchi, Claudio Gennaro
Publisher	Springer Verlag
Pages	339-344
Number of pages	6
ISBN (Print)	9783319250861
DOIs	https://doi.org/10.1007/978-3-319-25087-8_33
Publication status	Published - 2015
Externally published	Yes
Event	8th International Conference on Similarity Search and Applications, SISAP 2015 - Glasgow, United Kingdom Duration: 2015 Oct 12 → 2015 Oct 14

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9371
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	8th International Conference on Similarity Search and Applications, SISAP 2015
Country/Territory	United Kingdom
City	Glasgow
Period	15/10/12 → 15/10/14

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-319-25087-8_33

Cite this

Hara, K., Suzuki, I., Kobayashi, K., Fukumizu, K., & Radovanović, M. (2015). Reducing hubness for kernel regression. In R. Connor, G. Amato, F. Falchi, & C. Gennaro (Eds.), Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings (pp. 339-344). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9371). Springer Verlag. https://doi.org/10.1007/978-3-319-25087-8_33

Reducing hubness for kernel regression. / Hara, Kazuo; Suzuki, Ikumi; Kobayashi, Kei et al.
Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings. ed. / Richard Connor; Giuseppe Amato; Fabrizio Falchi; Claudio Gennaro. Springer Verlag, 2015. p. 339-344 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9371).

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

Hara, K, Suzuki, I, Kobayashi, K, Fukumizu, K & Radovanović, M 2015, Reducing hubness for kernel regression. in R Connor, G Amato, F Falchi & C Gennaro (eds), Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9371, Springer Verlag, pp. 339-344, 8th International Conference on Similarity Search and Applications, SISAP 2015, Glasgow, United Kingdom, 15/10/12. https://doi.org/10.1007/978-3-319-25087-8_33

Hara K, Suzuki I, Kobayashi K, Fukumizu K, Radovanović M. Reducing hubness for kernel regression. In Connor R, Amato G, Falchi F, Gennaro C, editors, Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings. Springer Verlag. 2015. p. 339-344. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-25087-8_33

Hara, Kazuo ; Suzuki, Ikumi ; Kobayashi, Kei et al. / Reducing hubness for kernel regression. Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings. editor / Richard Connor ; Giuseppe Amato ; Fabrizio Falchi ; Claudio Gennaro. Springer Verlag, 2015. pp. 339-344 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{dcab97eb918e40369e269e7cae15b923,

title = "Reducing hubness for kernel regression",

abstract = "In this paper, we point out that hubness—some samples in a high-dimensional dataset emerge as hubs that are similar to many other samples—influences the performance of kernel regression. Because the dimension of feature spaces induced by kernels is usually very high, hubness occurs, giving rise to the problem of multicollinearity, which is known as a cause of instability of regression results. We propose hubnessreduced kernels for kernel regression as an extension of a previous approach for kNN classification that reduces spatial centrality to eliminate hubness.",

author = "Kazuo Hara and Ikumi Suzuki and Kei Kobayashi and Kenji Fukumizu and Milo{\v s} Radovanovi{\'c}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2015.; 8th International Conference on Similarity Search and Applications, SISAP 2015 ; Conference date: 12-10-2015 Through 14-10-2015",

year = "2015",

doi = "10.1007/978-3-319-25087-8_33",

language = "English",

isbn = "9783319250861",

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

publisher = "Springer Verlag",

pages = "339--344",

editor = "Richard Connor and Giuseppe Amato and Fabrizio Falchi and Claudio Gennaro",

booktitle = "Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings",

}

TY - GEN

T1 - Reducing hubness for kernel regression

AU - Hara, Kazuo

AU - Suzuki, Ikumi

AU - Kobayashi, Kei

AU - Fukumizu, Kenji

AU - Radovanović, Miloš

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2015.

PY - 2015

Y1 - 2015

N2 - In this paper, we point out that hubness—some samples in a high-dimensional dataset emerge as hubs that are similar to many other samples—influences the performance of kernel regression. Because the dimension of feature spaces induced by kernels is usually very high, hubness occurs, giving rise to the problem of multicollinearity, which is known as a cause of instability of regression results. We propose hubnessreduced kernels for kernel regression as an extension of a previous approach for kNN classification that reduces spatial centrality to eliminate hubness.

AB - In this paper, we point out that hubness—some samples in a high-dimensional dataset emerge as hubs that are similar to many other samples—influences the performance of kernel regression. Because the dimension of feature spaces induced by kernels is usually very high, hubness occurs, giving rise to the problem of multicollinearity, which is known as a cause of instability of regression results. We propose hubnessreduced kernels for kernel regression as an extension of a previous approach for kNN classification that reduces spatial centrality to eliminate hubness.

UR - http://www.scopus.com/inward/record.url?scp=84951787726&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84951787726&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-25087-8_33

DO - 10.1007/978-3-319-25087-8_33

M3 - Conference contribution

AN - SCOPUS:84951787726

SN - 9783319250861

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

SP - 339

EP - 344

BT - Similarity Search and Applications - 8th International Conference, SISAP 2015, Proceedings

A2 - Connor, Richard

A2 - Amato, Giuseppe

A2 - Falchi, Fabrizio

A2 - Gennaro, Claudio

PB - Springer Verlag

T2 - 8th International Conference on Similarity Search and Applications, SISAP 2015

Y2 - 12 October 2015 through 14 October 2015

ER -

Reducing hubness for kernel regression

Abstract

Publication series

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this