Algorithmes de Bi-regroupement pour l’Analyse des Données Complexes.

Abstract

Biclustering, a well-known bioinformatics technique, is essential for analyzing gene expression data because it reveals patterns and identifies groupings of genes that behave similarly under particular conditions. This thesis aims to contribute to the field through the introduction of three distinct approaches: a Differential Evolution-based method, a Multi-Objective Differential Evolution-based approach featuring a novel adaptive mutation operator known as BBDE, and a final method that employs Convolutional Denoising Autoencoders (CDAs) for preprocessing followed by Artificial Bee Colony (ABC) for biclustering. Each strategy displays its usefulness via comprehensive findings, contributing to the progress of biclustering techniques and improving gene expression data analysis in computational genomics.