Robust orthogonal nonnegative matrix tri-factorization for data representation
Nonnegative matrix factorization (NMF) has been a vital data representation technique, and has demonstrated significant potential in the field of machine learning and data mining. Nonnegative matrix tri-factorization (NMTF) is an extension of NMF, and provides more degrees of freedom than NMF. In th...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Journal Article |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/161108 |
| _version_ | 1826123078152749056 |
|---|---|
| author | Peng, Siyuan Ser, Wee Chen, Badong Lin, Zhiping |
| author2 | School of Electrical and Electronic Engineering |
| author_facet | School of Electrical and Electronic Engineering Peng, Siyuan Ser, Wee Chen, Badong Lin, Zhiping |
| author_sort | Peng, Siyuan |
| collection | NTU |
| description | Nonnegative matrix factorization (NMF) has been a vital data representation technique, and has demonstrated significant potential in the field of machine learning and data mining. Nonnegative matrix tri-factorization (NMTF) is an extension of NMF, and provides more degrees of freedom than NMF. In this paper, we propose the correntropy based orthogonal nonnegative matrix tri-factorization (CNMTF) algorithm, which is robust to noisy data contaminated by non-Gaussian noise and outliers. Different from previous NMF algorithms, CNMTF firstly applies correntropy to NMTF to measure the similarity, and preserves double orthogonality conditions and dual graph regularization. We adopt the half-quadratic technique to solve the optimization problem of CNMTF, and derive the multiplicative update rules. The complexity issue of CNMTF is also presented. Furthermore, the robustness of the proposed algorithm is analyzed, and the relationships between CNMTF and several previous NMF based methods are discussed. Experimental results demonstrate that the proposed CNMTF method has better performance on real world image and text datasets for clustering tasks, compared with several state-of-the-art methods. |
| first_indexed | 2024-10-01T05:58:44Z |
| format | Journal Article |
| id | ntu-10356/161108 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:58:44Z |
| publishDate | 2022 |
| record_format | dspace |
| spelling | ntu-10356/1611082022-08-16T02:56:50Z Robust orthogonal nonnegative matrix tri-factorization for data representation Peng, Siyuan Ser, Wee Chen, Badong Lin, Zhiping School of Electrical and Electronic Engineering Engineering::Electrical and electronic engineering Correntropy Orthogonality Constraint Nonnegative matrix factorization (NMF) has been a vital data representation technique, and has demonstrated significant potential in the field of machine learning and data mining. Nonnegative matrix tri-factorization (NMTF) is an extension of NMF, and provides more degrees of freedom than NMF. In this paper, we propose the correntropy based orthogonal nonnegative matrix tri-factorization (CNMTF) algorithm, which is robust to noisy data contaminated by non-Gaussian noise and outliers. Different from previous NMF algorithms, CNMTF firstly applies correntropy to NMTF to measure the similarity, and preserves double orthogonality conditions and dual graph regularization. We adopt the half-quadratic technique to solve the optimization problem of CNMTF, and derive the multiplicative update rules. The complexity issue of CNMTF is also presented. Furthermore, the robustness of the proposed algorithm is analyzed, and the relationships between CNMTF and several previous NMF based methods are discussed. Experimental results demonstrate that the proposed CNMTF method has better performance on real world image and text datasets for clustering tasks, compared with several state-of-the-art methods. Nanyang Technological University This work was partially supported by Nanyang Technological University Research Scholarships of Singapore, National Basic Research Program (973 Program) of China (No. 2015CB351703), and National Natural Science Foundation of China (No. 91648208). 2022-08-16T02:56:50Z 2022-08-16T02:56:50Z 2020 Journal Article Peng, S., Ser, W., Chen, B. & Lin, Z. (2020). Robust orthogonal nonnegative matrix tri-factorization for data representation. Knowledge-Based Systems, 201-202, 106054-. https://dx.doi.org/10.1016/j.knosys.2020.106054 0950-7051 https://hdl.handle.net/10356/161108 10.1016/j.knosys.2020.106054 2-s2.0-85085636981 201-202 106054 en Knowledge-Based Systems © 2020 Elsevier B.V. All rights reserved. |
| spellingShingle | Engineering::Electrical and electronic engineering Correntropy Orthogonality Constraint Peng, Siyuan Ser, Wee Chen, Badong Lin, Zhiping Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title | Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title_full | Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title_fullStr | Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title_full_unstemmed | Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title_short | Robust orthogonal nonnegative matrix tri-factorization for data representation |
| title_sort | robust orthogonal nonnegative matrix tri factorization for data representation |
| topic | Engineering::Electrical and electronic engineering Correntropy Orthogonality Constraint |
| url | https://hdl.handle.net/10356/161108 |
| work_keys_str_mv | AT pengsiyuan robustorthogonalnonnegativematrixtrifactorizationfordatarepresentation AT serwee robustorthogonalnonnegativematrixtrifactorizationfordatarepresentation AT chenbadong robustorthogonalnonnegativematrixtrifactorizationfordatarepresentation AT linzhiping robustorthogonalnonnegativematrixtrifactorizationfordatarepresentation |