Robust non-negative matrix factorization for subspace learning

Authors

  • Xiangguang Dai Chongqing Three Gorges University - School of Three Gorges Artificial Intelligence
  • Yingyin Tao Chongqing Three Gorges University - School of Three Gorges Artificial Intelligence
  • Wei Zhang Chongqing Three Gorges University - Engineering Research Center of Internet of Things and Intelligent Control
  • Yuming Feng Chongqing Three Gorges University - Key Laboratory of Intelligent Information Processing and Control

Keywords:

non-negative matrix factorization, matrix completion, robustness, clustering

Abstract

Conventional non-negative matrix factorization methods fail to clusetring due to many noisy data.  Therefore, this paper proposes a novel non-negative matrix factorization method to learn a robust subspace from the noisy space.  Considering matrix completion and non-negative matrix factorization, the pro-posed method can recover the contaminated data and learn a low-dimensional effective subspace.  Experiments on the ORL face dataset with Salt & Pepper noise show that our proposed method is effective and robust.

Downloads

Published

2020-07-31

How to Cite

Dai, X., Tao, Y., Zhang, W., & Feng, Y. (2020). Robust non-negative matrix factorization for subspace learning. Italian Journal of Pure and Applied Mathematics, 44, 511–520. Retrieved from https://journals.uniurb.it/index.php/ijpam/article/view/7454

Issue

Section

Articoli - Forum Editrice

Similar Articles

1 2 3 4 5 6 > >> 

You may also start an advanced similarity search for this article.