Human mental states classification using EEG by means of Genetic Programming

Emigdio Z-Flores 1
1 CQFD - Quality control and dynamic reliability
IMB - Institut de Mathématiques de Bordeaux, Inria Bordeaux - Sud-Ouest
Résumé : The advances in the development of Brain-Computer Interfaces (BCI) have been increasing in recent years, mostly because the level of convergence from multi-disciplinary techniques has evolved. The electroencephalography (EEG), a brain recording method studied in this thesis, allows the construction of BCIs, however the signals are rather complex to process, which requires methodologies that efficiently extract patterns from them. This thesis explores two directions: first, a system is proposed for the epilepsy seizures recognition using a combination of signal processing methods for an efficient feature extraction; second, it explores the usage of a meta-heuristic algorithm, namely Genetic Programming (GP), as an alternative in the design of BCIs. Nonetheless, there is currently open-issues in GP that this thesis also explores: is there a more efficient search methodology in the exploration by GP?; what is a proper representation depending on the studied problem?; which are the most adequate search operators?. For the first topic, a thoroughly study is presented by introducing a memetic GP applied to regression problems. Then, it is extended by adapting it to classification problems. The results are positive; GP is greatly benefited from the combination of a general and a Local Search (LS) methodology. The last two topics are studied simultaneously in the development of a recognition system for mental states using EEG. A GP version (+FEGP) is proposed that evolves feature extraction models by using specialized search operators, individuals representation and fitness function. The results show that the combination of these reaches a state-of-the-art accuracy for the particular task of mental states recognition.
Liste complète des métadonnées

Littérature citée [352 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/tel-01668672
Contributeur : Pierrick Legrand <>
Soumis le : mercredi 20 décembre 2017 - 11:04:25
Dernière modification le : jeudi 11 janvier 2018 - 06:22:12

Fichier

PhD_Thesis_Emigdio.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : tel-01668672, version 1

Collections

Citation

Emigdio Z-Flores. Human mental states classification using EEG by means of Genetic Programming. Artificial Intelligence [cs.AI]. ITT, Instituto tecnologico de Tijuana, 2017. English. 〈tel-01668672〉

Partager

Métriques

Consultations de la notice

156

Téléchargements de fichiers

68