Abstract
Matrix rank minimizing subject to affine constraints arises in many application areas, ranging from signal processing to machine learning. Nuclear norm is a convex relaxation for this problem which can recover the rank exactly under some restricted and theoretically interesting conditions. However, for many real-world applications, nuclear norm approximation to the rank function can only produce a result far from the optimum. To seek a solution of higher accuracy than the nuclear norm, in this letter, we propose a rank approximation based on Logarithm-Determinant. We consider using this rank approximation for subspace clustering application. Our framework can model different kinds of errors and noise. Effective optimization strategy is developed with theoretical guarantee to converge to a stationary point. The proposed method gives promising results on face clustering and motion segmentation tasks compared to the state-of-the-art subspace clustering algorithms.
| Original language | English |
|---|---|
| Article number | 7166307 |
| Pages (from-to) | 2088-2092 |
| Number of pages | 5 |
| Journal | IEEE Signal Processing Letters |
| Volume | 22 |
| Issue number | 11 |
| DOIs | |
| State | Published - Nov 1 2015 |
Bibliographical note
Publisher Copyright:© 2015 IEEE.
Funding
| Funders | Funder number |
|---|---|
| U.S. Department of Energy Chinese Academy of Sciences Guangzhou Municipal Science and Technology Project Oak Ridge National Laboratory Extreme Science and Engineering Discovery Environment National Science Foundation National Energy Research Scientific Computing Center National Natural Science Foundation of China | 1218712 |
Keywords
- Matrix rank minimization
- nonconvex optimization
- nuclear norm
- subspace clustering
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering
- Applied Mathematics
Fingerprint
Dive into the research topics of 'Robust Subspace Clustering via Smoothed Rank Approximation'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver