Convex Algorithms for Nonnegative Matrix Factorization

TL;DR

This paper introduces convex approximation algorithms for nonnegative matrix factorization, enabling efficient solutions to a traditionally challenging problem with applications demonstrated on numerical examples.

Contribution

It presents novel convex approximation algorithms for NMF, improving computational efficiency and solution quality over existing methods.

Findings

Algorithms successfully factorize matrices in tested examples

Convex approximations outperform some traditional methods

Efficient solutions with potential for broader applications

Abstract

We derive approximation algorithms for the nonnegative matrix factorization problem, i.e. the problem of factorizing a matrix as the product of two matrices with nonnegative coefficients. We form convex approximations of this problem which can be solved efficiently and test our algorithms on some classic numerical examples.