Nonnegative matrix factorization on orthogonal subspace with smoothed l0 norm constrained

Authors:
Jun Ye;Zhong Jin
Affiliations:
School of Computer Science & Technology, Nanjing University of Science and Technology, Nanjing, China,School of Natural Sciences, Nanjing University of Posts & Telecommunications, Nanjing, ...;School of Computer Science & Technology, Nanjing University of Science and Technology, Nanjing, China
Venue:
IScIDE'12 Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
Year:
2012

Citing 8
Cited 0

Non-negative Matrix Factorization with Sparseness Constraints

The Journal of Machine Learning Research
Orthogonal nonnegative matrix t-factorizations for clustering

Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining
Nonnegative matrix factorization with constrained second-order optimization

Signal Processing
Multiplicative updates for non-negative projections

Neurocomputing
A fast approach for overcomplete sparse decomposition based on smoothed l0 norm

IEEE Transactions on Signal Processing
Nonnegative Matrix Factorization on Orthogonal Subspace

Pattern Recognition Letters
Linear and nonlinear projective nonnegative matrix factorization

IEEE Transactions on Neural Networks
Stable recovery of sparse overcomplete representations in the presence of noise

IEEE Transactions on Information Theory

Quantified Score

Hi-index	0.00

Visualization

Abstract

It is known that the sparseness of the factor matrices by Nonnegative Matrix Factorization can influence the clustering performance. In order to improve the ability of the sparse representations of the NMF, we proposed the new algorithm for Nonnegatie Matrix Factorization, coined nonnegative matrix factorization on orthogonal subspace with smoothed L0 norm constrained, in which the generation of orthogonal factor matrices with smoothed L0 norm constrained are the parts of objective function minimization. Also we develop simple multiplicative updates for our proposed method. Experiment on three real-world databases (Iris, UCI, ORL) show that our proposed method can achieve the best or close to the best in clustering and in the way of the sparse representation than other methods.