LE PELTIER, Mme Clarisse (2020) Réduire le langage à un processus : Learning ou Deep Learning ? PRE - Research Project, ENSTA.



We can speak, this may seem a little bit like stating the obvious but can we explain properly the way we learn to speak ? Beyond vocabulary, we understand each other thanks to grammar which allows us to construct sentences, order our words and make them meaningful. An important dimension in grammar is our syntax (words ordering, grammatical types, syntactic meaning...) In the early months of life babies assimilate a large number of process that help them quickly understand their environment. Long before they can speak babies identify the main grammatical types without knowing how to speak. For a long time already researchers are particularly interested in this grammatical types learning process thanks to many years of experiences (some of wich are monitored by EEG-measures). This work aims at considering the possibility of treating EEG data with automated classification ( i.e Deep Learning). Such classification would be carried out by convolutionnal neural networks (CNN) as they can possibly detect patterns in signals. More precisely the specific signals of nouns and verb treatment will be the heart of the matter.

Item Type:Thesis (PRE - Research Project)
Uncontrolled Keywords:Langage, Réseaux de neurones, EEG, Classification, Apprentissage,
Subjects:Information and Communication Sciences and Technologies
Life Sciences and Engineering
ID Code:8163
Deposited By:Clarisse LE PELTIER
Deposited On:31 août 2020 16:21
Dernière modification:31 août 2020 16:21

Repository Staff Only: item control page