Mod\`ele \`a processus latent et algorithme EM pour la r\'egression non lin\'eaire

TL;DR

This paper introduces a non-linear regression model with a latent process, estimated using an EM algorithm, demonstrating effective performance on simulated and real data.

Contribution

It presents a novel non-linear regression model incorporating a latent process and an EM algorithm for parameter estimation.

Findings

Good performance on simulated data

Effective on real data sets

Smooth activation of polynomial regression models

Abstract

A non linear regression approach which consists of a specific regression model incorporating a latent process, allowing various polynomial regression models to be activated preferentially and smoothly, is introduced in this paper. The model parameters are estimated by maximum likelihood performed via a dedicated expecation-maximization (EM) algorithm. An experimental study using simulated and real data sets reveals good performances of the proposed approach.