Damien Lesens

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

This work argues that the KL-NMF method has reached its limits and proposes to use instead the second-order Taylor expansion of the loss, leading to a Newton-type method which provably converges and competes favorably with state-of-the-art algorithms on a large variety of datasets.