Description
Date depot: 29 octobre 2019
Titre: Contextualized and Personalized Neural Networks
Directeur de thèse:
Matthieu CORD (ISIR (EDITE))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Intelligence artificielle
Resumé:
Deep-Learning has made significant progress in the last years in different fields, such as Computer Vision, Natural Language Processing; etc. Nevertheless, there is still a lot to be done: on the one hand, current systems are not yet capable of accomplishing tasks as complex as the ones that can be performed by humans: indeed, the human being is capable of performing different tasks with different granularity, adapt it to their interlocutor, to the context, etc.
The objective of this PhD project is to study theoretically and implement contextualized and personalized Neural Networks. Particularly, we will design systems that allow the neural network to have an attribute-based definition of the tasks defined by the user (possibly with interactive learning), being able to integrate other data sources, like text associated with the image or could take into account the user intent or a user profile, that would be a way to personalize the outcome of the network.
Doctorant.e: Touvron Hugo