Jérémy E. Cohen

An Efficient Newton Algorithm for Nonnegative Matrix Factorization with the Kullback-Leibler Divergence

Damien Lesens, Jérémy E. Cohen, Bora Uçar

Nonnegative Matrix Factorization (NMF) is a fundamental tool in unsupervised learning, which approximates a nonnegative matrix by the product of two low-rank nonnegative factors. The Kullback-Leibler (KL) divergence is best suited to measure the data to model discrepancy when the…Read more