A new inequality for maximum likelihood estimation in statistical models with latent variables

TL;DR

This paper introduces a new inequality involving posterior distributions in latent variable models, which generalizes existing methods and has potential applications similar to the EM algorithm.

Contribution

The paper presents a novel inequality for posterior distributions in latent variable models, expanding theoretical tools for maximum likelihood estimation.

Findings

Provides a new inequality applicable under general conditions

Links the inequality to the EM algorithm framework

Potential to improve MLE methods in latent variable models

Abstract

Maximum-likelihood estimation (MLE) is arguably the most important tool for statisticians, and many methods have been developed to find the MLE. We present a new inequality involving posterior distributions of a latent variable that holds under very general conditions. It is related to the EM algorithm and has a clear potential for being used in a similar fashion.