Keywords: Divergence, Robust Statistics, Clustering
TL;DR: This paper introduces the $t$-divergence using the inverse tangent function, highlighting its robustness, quasi-metric properties, and utility in high-dimensional data clustering.
Abstract: This paper introduces the $t$-divergence, a novel divergence measure associated with the inverse tangent function. We investigate its intriguing consistent and outlier-robust features, particularly its quasi-metric properties and role in establishing weak convergence. Additionally, we showcase the efficacy of this divergence measure family in feature-weighted clustering for high-dimensional data.
Submission Number: 81
Loading