Hyperprior on symmetric Dirichlet distribution

TL;DR

This paper introduces a method to place vague hyperpriors on Dirichlet distributions, updates parameters using adaptive rejection sampling, and analyzes the approach within over-fitted mixture models through synthetic experiments.

Contribution

It presents a novel approach for hyperprior placement on Dirichlet distributions and demonstrates its effectiveness in over-fitted mixture models.

Findings

Hyperprior improves model flexibility

ARS effectively updates Dirichlet parameters

Synthetic experiments validate the approach

Abstract

In this article we introduce how to put vague hyperprior on Dirichlet distribution, and we update the parameter of it by adaptive rejection sampling (ARS). Finally we analyze this hyperprior in an over-fitted mixture model by some synthetic experiments.