Bayesian SPLDA

TL;DR

This paper derives equations for Variational Bayes estimation of SPLDA parameters, enabling adaptation with limited data and a fully Bayesian approach, similar to Bishop's VB PPCA.

Contribution

It introduces a Variational Bayes framework for SPLDA, allowing efficient parameter estimation and adaptation with minimal data, extending existing probabilistic models.

Findings

Derived equations for VB SPLDA estimation

Enables adaptation with limited data

Provides a fully Bayesian implementation

Abstract

In this document we are going to derive the equations needed to implement a Variational Bayes estimation of the parameters of the simplified probabilistic linear discriminant analysis (SPLDA) model. This can be used to adapt SPLDA from one database to another with few development data or to implement the fully Bayesian recipe. Our approach is similar to Bishop's VB PPCA.