English

Generalized Separable Nonnegative Matrix Factorization

Machine Learning 2021-04-14 v2 Computer Vision and Pattern Recognition Optimization and Control Machine Learning

Abstract

Nonnegative matrix factorization (NMF) is a linear dimensionality technique for nonnegative data with applications such as image analysis, text mining, audio source separation and hyperspectral unmixing. Given a data matrix MM and a factorization rank rr, NMF looks for a nonnegative matrix WW with rr columns and a nonnegative matrix HH with rr rows such that MWHM \approx WH. NMF is NP-hard to solve in general. However, it can be computed efficiently under the separability assumption which requires that the basis vectors appear as data points, that is, that there exists an index set K\mathcal{K} such that W=M(:,K)W = M(:,\mathcal{K}). In this paper, we generalize the separability assumption: We only require that for each rank-one factor W(:,k)H(k,:)W(:,k)H(k,:) for k=1,2,,rk=1,2,\dots,r, either W(:,k)=M(:,j)W(:,k) = M(:,j) for some jj or H(k,:)=M(i,:)H(k,:) = M(i,:) for some ii. We refer to the corresponding problem as generalized separable NMF (GS-NMF). We discuss some properties of GS-NMF and propose a convex optimization model which we solve using a fast gradient method. We also propose a heuristic algorithm inspired by the successive projection algorithm. To verify the effectiveness of our methods, we compare them with several state-of-the-art separable NMF algorithms on synthetic, document and image data sets.

Keywords

Cite

@article{arxiv.1905.12995,
  title  = {Generalized Separable Nonnegative Matrix Factorization},
  author = {Junjun Pan and Nicolas Gillis},
  journal= {arXiv preprint arXiv:1905.12995},
  year   = {2021}
}

Comments

31 pages, 12 figures, 4 tables. We have added discussions about the identifiability of the model, we have modified the first synthetic experiment, we have clarified some aspects of the contribution