Nonnegative Matrix Factorization Based On Class Information And Sparse Representation

Nonnegative matrix factorization get the linear representation of the basis matrix.This method has been widely used in pattern recognition,and information retrieval.For pattern recognition problems,basic nonnegative matrix decomposition method does not consider the category information.Besides,sparse representation has higher recognition rate in pattern recognition problems.So the category information and sparse representation should be used to solve the pattern recognition problems,the experimental result proves that this method has higher pattern recognition rate.