Skip to Main content Skip to Navigation
Theses

Logique floue et algorithmes génétiques pour le pré-traitement de données de biopuces et la sélection de gènes

Abstract : In molecular biology, technologies for gene expression analyses, as DNA microarrays, give rise to a lot of research. One possible application of such technologies is diagnosis and recognition of different kinds of tumours. One particularity of microarray data is their great number of attributes (genes) whereas very few samples are available. This dimensionality problem makes the data difficult to understand and reduces the efficiency of classification algorithms. This thesis proposes newmethods to reduce the initial dimension of data and to select relevant gene subsets for classification. First, we propose a method for data pre-processing and dimension reduction based on fuzzy logic. Then the problem of gene selection is treated by two kinds of approaches. In the first one, we propose a wrapper method that selects a relevant gene subset for a SVM classifier by a double genetic exploration. In the second approach, we propose an embedded method where LDA classifier provides information about gene relevancy to guide the genetic algorithm. The different experimentations that we have performed give very good results, especially for the embedded method.
Document type :
Theses
Complete list of metadatas

Cited literature [126 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00451112
Contributor : Anne-Marie Plé <>
Submitted on : Thursday, January 28, 2010 - 11:48:47 AM
Last modification on : Monday, June 1, 2020 - 4:54:02 PM
Long-term archiving on: : Friday, June 18, 2010 - 5:00:49 PM

Identifiers

  • HAL Id : tel-00451112, version 1

Citation

Edmundo Bonilla Huerta. Logique floue et algorithmes génétiques pour le pré-traitement de données de biopuces et la sélection de gènes. Informatique [cs]. Université d'Angers, 2008. Français. ⟨tel-00451112⟩

Share

Metrics

Record views

595

Files downloads

5623