Biclustering multivariate data for correlated subspace mining

Kazuho Watanabe, Hsiang Yun Wu, Yusuke Niibe, Shigeo Takahashi, Issei Fujishiro

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

7 Citations (Scopus)

Abstract

Exploring feature subspaces is one of promising approaches to analyzing and understanding the important patterns in multivariate data. If relying too much on effective enhancements in manual interventions, the associated results depend heavily on the knowledge and skills of users performing the data analysis. This paper presents a novel approach to extracting feature subspaces from multivariate data by incorporating biclustering techniques. The approach has been maximally automated in the sense that highly-correlated dimensions are automatically grouped to form subspaces, which effectively supports further exploration of them. A key idea behind our approach lies in a new mathematical formulation of asymmetric biclustering, by combining spherical k-means clustering for grouping highly-correlated dimensions, together with ordinary k-means clustering for identifying subsets of data samples. Lower-dimensional representations of data in feature subspaces are successfully visualized by parallel coordinate plot, where we project the data samples of correlated dimensions to one composite axis through dimensionality reduction schemes. Several experimental results of our data analysis together with discussions will be provided to assess the capability of our approach.

Original languageEnglish
Title of host publicationIEEE Pacific Visualization Symposium
PublisherIEEE Computer Society
Pages287-294
Number of pages8
Volume2015-July
ISBN (Print)9781467368797
DOIs
Publication statusPublished - 2015 Jul 14
Event2015 8th IEEE Pacific Visualization Symposium, PacificVis 2015 - Hangzhou, China
Duration: 2015 Apr 142015 Apr 17

Other

Other2015 8th IEEE Pacific Visualization Symposium, PacificVis 2015
CountryChina
CityHangzhou
Period15/4/1415/4/17

Fingerprint

Composite materials

Keywords

  • biclustering
  • correlation
  • Multivariate data
  • subspaces

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition
  • Hardware and Architecture
  • Software

Cite this

Watanabe, K., Wu, H. Y., Niibe, Y., Takahashi, S., & Fujishiro, I. (2015). Biclustering multivariate data for correlated subspace mining. In IEEE Pacific Visualization Symposium (Vol. 2015-July, pp. 287-294). [7156389] IEEE Computer Society. https://doi.org/10.1109/PACIFICVIS.2015.7156389

Biclustering multivariate data for correlated subspace mining. / Watanabe, Kazuho; Wu, Hsiang Yun; Niibe, Yusuke; Takahashi, Shigeo; Fujishiro, Issei.

IEEE Pacific Visualization Symposium. Vol. 2015-July IEEE Computer Society, 2015. p. 287-294 7156389.

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

Watanabe, K, Wu, HY, Niibe, Y, Takahashi, S & Fujishiro, I 2015, Biclustering multivariate data for correlated subspace mining. in IEEE Pacific Visualization Symposium. vol. 2015-July, 7156389, IEEE Computer Society, pp. 287-294, 2015 8th IEEE Pacific Visualization Symposium, PacificVis 2015, Hangzhou, China, 15/4/14. https://doi.org/10.1109/PACIFICVIS.2015.7156389
Watanabe K, Wu HY, Niibe Y, Takahashi S, Fujishiro I. Biclustering multivariate data for correlated subspace mining. In IEEE Pacific Visualization Symposium. Vol. 2015-July. IEEE Computer Society. 2015. p. 287-294. 7156389 https://doi.org/10.1109/PACIFICVIS.2015.7156389
Watanabe, Kazuho ; Wu, Hsiang Yun ; Niibe, Yusuke ; Takahashi, Shigeo ; Fujishiro, Issei. / Biclustering multivariate data for correlated subspace mining. IEEE Pacific Visualization Symposium. Vol. 2015-July IEEE Computer Society, 2015. pp. 287-294
@inproceedings{cc32139fc8734b579afac1f838e6d777,
title = "Biclustering multivariate data for correlated subspace mining",
abstract = "Exploring feature subspaces is one of promising approaches to analyzing and understanding the important patterns in multivariate data. If relying too much on effective enhancements in manual interventions, the associated results depend heavily on the knowledge and skills of users performing the data analysis. This paper presents a novel approach to extracting feature subspaces from multivariate data by incorporating biclustering techniques. The approach has been maximally automated in the sense that highly-correlated dimensions are automatically grouped to form subspaces, which effectively supports further exploration of them. A key idea behind our approach lies in a new mathematical formulation of asymmetric biclustering, by combining spherical k-means clustering for grouping highly-correlated dimensions, together with ordinary k-means clustering for identifying subsets of data samples. Lower-dimensional representations of data in feature subspaces are successfully visualized by parallel coordinate plot, where we project the data samples of correlated dimensions to one composite axis through dimensionality reduction schemes. Several experimental results of our data analysis together with discussions will be provided to assess the capability of our approach.",
keywords = "biclustering, correlation, Multivariate data, subspaces",
author = "Kazuho Watanabe and Wu, {Hsiang Yun} and Yusuke Niibe and Shigeo Takahashi and Issei Fujishiro",
year = "2015",
month = "7",
day = "14",
doi = "10.1109/PACIFICVIS.2015.7156389",
language = "English",
isbn = "9781467368797",
volume = "2015-July",
pages = "287--294",
booktitle = "IEEE Pacific Visualization Symposium",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Biclustering multivariate data for correlated subspace mining

AU - Watanabe, Kazuho

AU - Wu, Hsiang Yun

AU - Niibe, Yusuke

AU - Takahashi, Shigeo

AU - Fujishiro, Issei

PY - 2015/7/14

Y1 - 2015/7/14

N2 - Exploring feature subspaces is one of promising approaches to analyzing and understanding the important patterns in multivariate data. If relying too much on effective enhancements in manual interventions, the associated results depend heavily on the knowledge and skills of users performing the data analysis. This paper presents a novel approach to extracting feature subspaces from multivariate data by incorporating biclustering techniques. The approach has been maximally automated in the sense that highly-correlated dimensions are automatically grouped to form subspaces, which effectively supports further exploration of them. A key idea behind our approach lies in a new mathematical formulation of asymmetric biclustering, by combining spherical k-means clustering for grouping highly-correlated dimensions, together with ordinary k-means clustering for identifying subsets of data samples. Lower-dimensional representations of data in feature subspaces are successfully visualized by parallel coordinate plot, where we project the data samples of correlated dimensions to one composite axis through dimensionality reduction schemes. Several experimental results of our data analysis together with discussions will be provided to assess the capability of our approach.

AB - Exploring feature subspaces is one of promising approaches to analyzing and understanding the important patterns in multivariate data. If relying too much on effective enhancements in manual interventions, the associated results depend heavily on the knowledge and skills of users performing the data analysis. This paper presents a novel approach to extracting feature subspaces from multivariate data by incorporating biclustering techniques. The approach has been maximally automated in the sense that highly-correlated dimensions are automatically grouped to form subspaces, which effectively supports further exploration of them. A key idea behind our approach lies in a new mathematical formulation of asymmetric biclustering, by combining spherical k-means clustering for grouping highly-correlated dimensions, together with ordinary k-means clustering for identifying subsets of data samples. Lower-dimensional representations of data in feature subspaces are successfully visualized by parallel coordinate plot, where we project the data samples of correlated dimensions to one composite axis through dimensionality reduction schemes. Several experimental results of our data analysis together with discussions will be provided to assess the capability of our approach.

KW - biclustering

KW - correlation

KW - Multivariate data

KW - subspaces

UR - http://www.scopus.com/inward/record.url?scp=84942238914&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84942238914&partnerID=8YFLogxK

U2 - 10.1109/PACIFICVIS.2015.7156389

DO - 10.1109/PACIFICVIS.2015.7156389

M3 - Conference contribution

AN - SCOPUS:84942238914

SN - 9781467368797

VL - 2015-July

SP - 287

EP - 294

BT - IEEE Pacific Visualization Symposium

PB - IEEE Computer Society

ER -