Séminaire (Organisé par l’Equipe de recherche DI)

Ludovic DENOYER

Maître de conférences, Université Pierre et Marie Curie - LIP6


Sequential Model for Budgeted Classification


Mardi 5 novembre 2013 à 14h en salle A108


Résumé :

With the emergence of Big Data, the conventional methods of learning are no longer appropriate, first because they have not been developed to take into account application constraints (time, memory usage, ...) and second because they are mainly focused on the treatment of simple data while Web data often have a complex topology.

The talk will describe how these two aspects can be handled by using sequential learning models (based on reinforcement learning techniques) for the problem of automatic classification and I will present different applications of these methods to real-world problems.



Actualités
Vidéothèque
Téléchargements
Annuaire



FR SHIC 3272

Collegium UTC/CNRS