Extracting Feature Subspace Using Modified Nonnegative Matrix Factorization

Non-negative matrix factorization (NMF) is useful in finding basis information of non-negative data. It is a new dimension reduction method. In this paper, we modified the original nonnegative matrix factorization in order to extract many basis vectors for each sample cluster. The primary idea is to extend the original NMF through incorporating the latent semantic space constraints inside the NMF decomposition. Finally, experimental results are given.