Ensemble estimation of multivariate f-divergence

Abstract

f-divergence estimation is an important problem in the fields of information theory, machine learning, and statistics. While several divergence estimators exist, relatively few of their convergence rates are known. We derive the MSE convergence rate for a density plug-in estimator of f-divergence. Then by applying the theory of optimally weighted ensemble estimation, we derive a divergence estimator with a convergence rate of O(1/T) that is simple to implement and performs well in high dimensions. We validate our theoretical results with experiments.

14 pages, 6 figures, a condensed version of this paper was accepted to ISIT 2014, Version 2: Moved the proofs of the theorems from the main body to appendices at the end

References in corpus (1)

• Ensemble estimation of multivariate f-divergence

Cited by in corpus (6)

• Efficient Estimation of Mutual Information for Strongly Dependent Variables
• Ensemble estimation of multivariate f-divergence
• Multivariate f-Divergence Estimation With Confidence
• Improving Zero-shot Voice Style Transfer via Disentangled Representation Learning
• Ensemble Estimation of Distributional Functionals via $k$-Nearest Neighbors
• Minimax Optimal Estimation of KL Divergence for Continuous Distributions