Orthogonal Nonnegative Matrix Factorization with Sparsity Constraints

TL;DR

This paper introduces a novel method for sparsity-constrained orthogonal nonnegative matrix factorization that effectively balances accuracy and constraints, using a reformulation as a facility location problem and a control-barrier function framework.

Contribution

It reformulates SCONMF as a capacity-constrained facility location problem and integrates control-barrier functions to enforce constraints, also providing a way to determine the true rank of factorization matrices.

Findings

Achieves significantly lower reconstruction errors, up to 150 times smaller.

Strictly satisfies non-negativity, orthogonality, and sparsity constraints.

Outperforms existing methods in accuracy and constraint adherence.

Abstract

This article presents a novel approach to solving the sparsity-constrained Orthogonal Nonnegative Matrix Factorization (SCONMF) problem, which requires decomposing a non-negative data matrix into the product of two lower-rank non-negative matrices, X=WH, where the mixing matrix H has orthogonal rows HH^T=I, while also satisfying an upper bound on the number of nonzero elements in each row. By reformulating SCONMF as a capacity-constrained facility-location problem (CCFLP), the proposed method naturally integrates non-negativity, orthogonality, and sparsity constraints. Specifically, our approach integrates control-barrier function (CBF) based framework used for dynamic optimal control design problems with maximum-entropy-principle-based framework used for facility location problems to enforce these constraints while ensuring robust factorization. Additionally, this work introduces a quantitative approach for determining the ``true" rank of W or H, equivalent to the number of ``true" features - a critical aspect in ONMF applications where the number of features is unknown. Simulations on various datasets demonstrate significantly improved factorizations with low reconstruction errors (as small as by 150 times) while strictly satisfying all constraints, outperforming existing methods that struggle with balancing accuracy and constraint adherence.