Un upper bound for Kullback-Leibler divergence with a small number of outliers

Alexander Gofman, Mark Kelbert

Abstract


We establish a new upper bound for the Kullback-Leibler divergence
of two discrete probability distributions which are close in a sense that typically the ratio of probabilities is nearly one and the number of outliers is small.

Full Text:

PDF


ISSN: 1331-0623 (Print), 1848-8013 (Online)