Tetsuya Takabatake, Keisuke Yano
Abstract
This paper studies a specific class of statistical divergences for spectral densities of time series: the spectral -Rényi divergences, which include the Itakura-Saito divergence as a limiting case. The aim of this paper is to highlight both information-theoretic and statistical properties of spectral -Rényi divergences. We reveal the connection between the spectral -Rényi divergence and the -divergence in robust statistics, and a variational representation of the spectral -Rényi divergence. Inspired by these results suggesting "robustness" of spectral