Abstract
Multi-label problems arise in various domains such as multi-topic document categorization, protein function prediction, and automatic image annotation. One natural way to deal with such problems is to construct a binary classifier for each label, resulting in a set of independent binary classification problems. Since multiple labels share the same input space, and the semantics conveyed by different labels are usually correlated, it is essential to exploit the correlation information contained in different labels. In this paper, we consider a general framework for extracting shared structures in multi-label classification. In this framework, a common subspace is assumed to be shared among multiple labels. We show that the optimal solution to the proposed formulation can be obtained by solving a generalized eigenvalue problem, though the problem is nonconvex. For high-dimensional problems, direct computation of the solution is expensive, and we develop an efficient algorithm for this case. One appealing feature of the proposed framework is that it includes several well-known algorithms as special cases, thus elucidating their intrinsic relationships. We further show that the proposed framework can be extended to the kernel-induced feature space. We have conducted extensive experiments on multi-topic web page categorization and automatic gene expression pattern image annotation tasks, and results demonstrate the effectiveness of the proposed formulation in comparison with several representative algorithms.
| Original language | English (US) |
|---|---|
| Article number | 8 |
| Journal | ACM Transactions on Knowledge Discovery from Data |
| Volume | 4 |
| Issue number | 2 |
| DOIs | |
| State | Published - May 1 2010 |
Fingerprint
Keywords
- Gene expression pattern image annotation
- Kernel methods
- Least squares loss
- Multi-label classification
- Shared subspace
- Singular value decomposition
- Web page categorization
ASJC Scopus subject areas
- Computer Science(all)
Cite this
A shared-subspace learning framework for multi-label classification. / Ji, Shuiwang; Tang, Lei; Yu, Shipeng; Ye, Jieping.
In: ACM Transactions on Knowledge Discovery from Data, Vol. 4, No. 2, 8, 01.05.2010.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - A shared-subspace learning framework for multi-label classification
AU - Ji, Shuiwang
AU - Tang, Lei
AU - Yu, Shipeng
AU - Ye, Jieping
PY - 2010/5/1
Y1 - 2010/5/1
N2 - Multi-label problems arise in various domains such as multi-topic document categorization, protein function prediction, and automatic image annotation. One natural way to deal with such problems is to construct a binary classifier for each label, resulting in a set of independent binary classification problems. Since multiple labels share the same input space, and the semantics conveyed by different labels are usually correlated, it is essential to exploit the correlation information contained in different labels. In this paper, we consider a general framework for extracting shared structures in multi-label classification. In this framework, a common subspace is assumed to be shared among multiple labels. We show that the optimal solution to the proposed formulation can be obtained by solving a generalized eigenvalue problem, though the problem is nonconvex. For high-dimensional problems, direct computation of the solution is expensive, and we develop an efficient algorithm for this case. One appealing feature of the proposed framework is that it includes several well-known algorithms as special cases, thus elucidating their intrinsic relationships. We further show that the proposed framework can be extended to the kernel-induced feature space. We have conducted extensive experiments on multi-topic web page categorization and automatic gene expression pattern image annotation tasks, and results demonstrate the effectiveness of the proposed formulation in comparison with several representative algorithms.
AB - Multi-label problems arise in various domains such as multi-topic document categorization, protein function prediction, and automatic image annotation. One natural way to deal with such problems is to construct a binary classifier for each label, resulting in a set of independent binary classification problems. Since multiple labels share the same input space, and the semantics conveyed by different labels are usually correlated, it is essential to exploit the correlation information contained in different labels. In this paper, we consider a general framework for extracting shared structures in multi-label classification. In this framework, a common subspace is assumed to be shared among multiple labels. We show that the optimal solution to the proposed formulation can be obtained by solving a generalized eigenvalue problem, though the problem is nonconvex. For high-dimensional problems, direct computation of the solution is expensive, and we develop an efficient algorithm for this case. One appealing feature of the proposed framework is that it includes several well-known algorithms as special cases, thus elucidating their intrinsic relationships. We further show that the proposed framework can be extended to the kernel-induced feature space. We have conducted extensive experiments on multi-topic web page categorization and automatic gene expression pattern image annotation tasks, and results demonstrate the effectiveness of the proposed formulation in comparison with several representative algorithms.
KW - Gene expression pattern image annotation
KW - Kernel methods
KW - Least squares loss
KW - Multi-label classification
KW - Shared subspace
KW - Singular value decomposition
KW - Web page categorization
UR - http://www.scopus.com/inward/record.url?scp=77953216761&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77953216761&partnerID=8YFLogxK
U2 - 10.1145/1754428.1754431
DO - 10.1145/1754428.1754431
M3 - Article
VL - 4
JO - ACM Transactions on Knowledge Discovery from Data
JF - ACM Transactions on Knowledge Discovery from Data
SN - 1556-4681
IS - 2
M1 - 8
ER -