Multiplicative Updates for NMF with $\beta$-Divergences under Disjoint Equality Constraints

Abstract

Nonnegative matrix factorization (NMF) is the problem of approximating an input nonnegative matrix, $V$, as the product of two smaller nonnegative matrices, $W$ and $H$. In this paper, we introduce a general framework to design multiplicative updates (MU) for NMF based on $\beta$-divergences ($\beta$-NMF) with disjoint equality constraints, and with penalty terms in the objective function. By disjoint, we mean that each variable appears in at most one equality constraint. Our MU satisfy the set of constraints after each update of the variables during the optimization process, while guaranteeing that the objective function decreases monotonically. We showcase this framework on three NMF models, and show that it competes favorably the state of the art: (1)~$\beta$-NMF with sum-to-one constraints on the columns of $H$, (2) minimum-volume $\beta$-NMF with sum-to-one constraints on the columns of $W$, and (3) sparse $\beta$-NMF with ${\ell }_2$-norm constraints on the columns of $W$.

arXiv commentWe have corrected an error in Table 1 and in Section 2.3: the updates are not correct for $\beta \in (1,2)$; see the paper for more details. We have highlighted the changes in red color

References in corpus (8)

• On Tensors, Sparsity, and Nonnegative Factorizations
• Nonnegative Matrix Factorization for Signal and Data Analytics: Identifiability, Algorithms, and Applications
• Accelerated Multiplicative Updates and Hierarchical ALS Algorithms for Nonnegative Matrix Factorization
• Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization
• Generalized Canonical Polyadic Tensor Decomposition
• Blind Audio Source Separation with Minimum-Volume Beta-Divergence NMF
• Accelerating Nonnegative Matrix Factorization Algorithms using Extrapolation
• Simplex-Structured Matrix Factorization: Sparsity-based Identifiability and Provably Correct Algorithms

Cited by in corpus (5)

• Majorization-minimization for Sparse Nonnegative Matrix Factorization with the $\beta$-divergence
• Partial Identifiability for Nonnegative Matrix Factorization
• Deep Nonnegative Matrix Factorization with Beta Divergences
• Bounded Simplex-Structured Matrix Factorization: Algorithms, Identifiability and Applications
• Block Majorization Minimization with Extrapolation and Application to $\beta$-NMF