IEICE TRANSACTIONS on Information
Unsupervised Dimension Reduction via Least-Squares Quadratic Mutual Information
Summary :
The goal of dimension reduction is to represent high-dimensional data in a lower-dimensional subspace, while intrinsic properties of the original data are kept as much as possible. An important challenge in unsupervised dimension reduction is the choice of tuning parameters, because no supervised information is available and thus parameter selection tends to be subjective and heuristic. In this paper, we propose an information-theoretic approach to unsupervised dimension reduction that allows objective tuning parameter selection. We employ quadratic mutual information (QMI) as our information measure, which is known to be less sensitive to outliers than ordinary mutual information, and QMI is estimated analytically by a least-squares method in a computationally efficient way. Then, we provide an eigenvector-based efficient implementation for performing unsupervised dimension reduction based on the QMI estimator. The usefulness of the proposed method is demonstrated through experiments.
- Publication
- IEICE TRANSACTIONS on Information Vol.E97-D No.10 pp.2806-2809
- Publication Date
- 2014/10/01
- Publicized
- 2014/07/22
- Online ISSN
- 1745-1361
- DOI
- 10.1587/transinf.2014EDL8111
- Type of Manuscript
- LETTER
- Category
- Artificial Intelligence, Data Mining
Authors
Janya SAINUI
Tokyo Institute of Technology
Masashi SUGIYAMA
Tokyo Institute of Technology
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Janya SAINUI, Masashi SUGIYAMA, "Unsupervised Dimension Reduction via Least-Squares Quadratic Mutual Information" in IEICE TRANSACTIONS on Information,
vol. E97-D, no. 10, pp. 2806-2809, October 2014, doi: 10.1587/transinf.2014EDL8111.
Abstract: The goal of dimension reduction is to represent high-dimensional data in a lower-dimensional subspace, while intrinsic properties of the original data are kept as much as possible. An important challenge in unsupervised dimension reduction is the choice of tuning parameters, because no supervised information is available and thus parameter selection tends to be subjective and heuristic. In this paper, we propose an information-theoretic approach to unsupervised dimension reduction that allows objective tuning parameter selection. We employ quadratic mutual information (QMI) as our information measure, which is known to be less sensitive to outliers than ordinary mutual information, and QMI is estimated analytically by a least-squares method in a computationally efficient way. Then, we provide an eigenvector-based efficient implementation for performing unsupervised dimension reduction based on the QMI estimator. The usefulness of the proposed method is demonstrated through experiments.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2014EDL8111/_p
Copy
@ARTICLE{e97-d_10_2806,
author={Janya SAINUI, Masashi SUGIYAMA, },
journal={IEICE TRANSACTIONS on Information},
title={Unsupervised Dimension Reduction via Least-Squares Quadratic Mutual Information},
year={2014},
volume={E97-D},
number={10},
pages={2806-2809},
abstract={The goal of dimension reduction is to represent high-dimensional data in a lower-dimensional subspace, while intrinsic properties of the original data are kept as much as possible. An important challenge in unsupervised dimension reduction is the choice of tuning parameters, because no supervised information is available and thus parameter selection tends to be subjective and heuristic. In this paper, we propose an information-theoretic approach to unsupervised dimension reduction that allows objective tuning parameter selection. We employ quadratic mutual information (QMI) as our information measure, which is known to be less sensitive to outliers than ordinary mutual information, and QMI is estimated analytically by a least-squares method in a computationally efficient way. Then, we provide an eigenvector-based efficient implementation for performing unsupervised dimension reduction based on the QMI estimator. The usefulness of the proposed method is demonstrated through experiments.},
keywords={},
doi={10.1587/transinf.2014EDL8111},
ISSN={1745-1361},
month={October},}
Copy
TY - JOUR
TI - Unsupervised Dimension Reduction via Least-Squares Quadratic Mutual Information
T2 - IEICE TRANSACTIONS on Information
SP - 2806
EP - 2809
AU - Janya SAINUI
AU - Masashi SUGIYAMA
PY - 2014
DO - 10.1587/transinf.2014EDL8111
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E97-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2014
AB - The goal of dimension reduction is to represent high-dimensional data in a lower-dimensional subspace, while intrinsic properties of the original data are kept as much as possible. An important challenge in unsupervised dimension reduction is the choice of tuning parameters, because no supervised information is available and thus parameter selection tends to be subjective and heuristic. In this paper, we propose an information-theoretic approach to unsupervised dimension reduction that allows objective tuning parameter selection. We employ quadratic mutual information (QMI) as our information measure, which is known to be less sensitive to outliers than ordinary mutual information, and QMI is estimated analytically by a least-squares method in a computationally efficient way. Then, we provide an eigenvector-based efficient implementation for performing unsupervised dimension reduction based on the QMI estimator. The usefulness of the proposed method is demonstrated through experiments.
ER -
English
