Scoring Formulation for Multi-Condition Joint PLDA

TL;DR

This paper extends the joint PLDA model to handle multiple nuisance conditions, providing a method to compute likelihood ratios for scoring in more complex scenarios involving shared nuisance variables.

Contribution

It introduces a generalized scoring formulation for joint PLDA accommodating multiple nuisance conditions, expanding its applicability.

Findings

Derived EM and scoring formulas for multiple nuisance conditions

Demonstrated how to compute likelihood ratios in complex scenarios

Extended joint PLDA model to more realistic conditions

Abstract

The joint PLDA model, is a generalization of PLDA where the nuisance variable is no longer considered independent across samples, but potentially shared (tied) across samples that correspond to the same nuisance condition. The original work considered a single nuisance condition, deriving the EM and scoring formulas for this scenario. In this document, we show how to obtain likelihood ratios for scoring when multiple nuisance conditions are allowed in the model.