A1335
Title: Robust generalized Bayesian inference via divergences for von Mises-Fisher distribution
Authors: Tomoyuki Nakagawa - Meisei University (Japan) [presenting]
Sho Kazari - Tokyo University of Science (Japan)
Kouji Tahata - Tokyo University of Science (Japan)
Yasuhito Tsuruta - The University of Nagano (Japan)
Abstract: The aim is to consider Bayesian inference based on the robust divergences for von Mises-Fisher distribution. Handling outliers is a common challenge in statistics, as they can significantly influence estimation results, even in directional data. While numerous studies have developed robust estimation methods for directional data with outliers in the frequentist frameworks, there are a few robust estimations in the Bayesian frameworks. The generalized posterior is provided using robust divergence for von Mises-Fisher distribution, and its asymptotic properties and robustness are demonstrated. Furthermore, the posterior computation algorithm is developed by adopting the weighted Bayesian bootstrap.
