Modèle à processus latent et algorithme EM pour la régression non linéaire
This work addresses non-linear regression problems for data analysis, but it appears incremental as it builds on existing EM and latent process methods.
The paper tackles non-linear regression by introducing a latent process model that smoothly activates various polynomial regressions, with parameters estimated via a dedicated EM algorithm, and experimental results on simulated and real data show good performance.
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.