Research Article
Multiview Learning via Non-negative Matrix Factorization for Clustering Applications
@INPROCEEDINGS{10.1007/978-3-031-04245-4_31, author={Jiajia Chen and Ao Li and Jie Li and Yangwei Wang}, title={Multiview Learning via Non-negative Matrix Factorization for Clustering Applications}, proceedings={6GN for Future Wireless Networks. 4th EAI International Conference, 6GN 2021, Huizhou, China, October 30--31, 2021, Proceedings}, proceedings_a={6GN}, year={2022}, month={5}, keywords={Non-negative matrix factorization Multiview clustering Similarity learning Spectral clustering}, doi={10.1007/978-3-031-04245-4_31} }- Jiajia Chen
Ao Li
Jie Li
Yangwei Wang
Year: 2022
Multiview Learning via Non-negative Matrix Factorization for Clustering Applications
6GN
Springer
DOI: 10.1007/978-3-031-04245-4_31
Abstract
Multiview clustering is to more fully use the information between views to guide the division of data points, and multiview data is often accompanied by high-dimensionality. Since non-negative matrix factorization can effectively extract features while reducing dimensionality, this paper proposed a multi-view learning method based on non-negative matrix factorization. Compared with other NMF-based multiview learning methods, the proposed method has the following advantages: 1) graph regularization is added to traditional NMF to explore potential popular structures, so that the learned similarity graph contains more potential information. 2) A common graph learning strategy is designed to integrate hidden information from different views. 3) Put the NMF-based similarity graph learning and common graph learning strategies into a unified framework, and optimize the similarity graph and common graph at the same time, so that the two promote each other. Experiments on three public datasets show that the proposed method is more robust than the existing methods.
