Ressource documentaire

Reconnaissance d’objets, une problématique résolue ? / Has object vision been solved? (en Anglais)


URL d'accès : http://www.canal-u.tv/?redirectVideo=23890...

Droits : © Inria Bordeaux - Sud-Ouest

Auteur(s) : THORPE Simon
10-07-2015

Description : The ability of humans to identify and categorize objects in complex natural scenes has long been thought to be beyond the capacities of artificial vision systems. However, recent progress in Deep Learning and Convolutional Neural Networks has demonstrated that simple feed-forward processing architectures composed of less than 10 layers of neurons can achieve human levels of performance in object recognition tasks. It is interesting to note that such processing architectures have a very similar structure to the primate visual system. Could it be that we are close to understanding how our brains recognize stimuli? I will argue that the main problem with the current state of the art in computer vision is that the learning procedures used are totally unrealistic. Essentially, building such a system requires hundreds of millions of training cycles of supervised learning. By contrast, our own visual systems can learn new stimuli in a few tens of presentations. I will suggest that more biologically realistic learning mechanisms based on spike-based processing and Spike Time Dependent Plasticity (STDP) may be much closer to the way our own visual systems operate, and allow our visual systems to learn about objects in the visual world on the basis of experience.
Mots-clés libres : neurosciences,information visuelle,extraction valeur des images
TECHNIQUE

Type : image en mouvement
Format : video/x-flv


Source(s) : 
rtmpt://fms2.cerimes.fr:80/vod/fuscia/reconnaissance.d.objets.une.problematique.resolue.has.object.vision.been.solved._23890/interfaces_simon.thorpe._compresse.e.mov


Entrepôt d'origine : Canal-u.fr
Identifiant : oai:canal-u.fr:23890
Type de ressource : Ressource documentaire
Exporter au format XML

Ressource pédagogique

Reconnaissance d’objets, une problématique résolue ? / Has object vision been solved? (en Anglais)


URL d'accès : http://www.canal-u.tv/video/inria/reconnaissance_d...
rtmpt://fms2.cerimes.fr:80/vod/fuscia/reconnaissan...
http://www.canal-u.tv/video/inria/dl.1/reconnaissa...

Identifiant de la fiche : 23890
Schéma de la métadonnée : LOMv1.0, LOMFRv1.0

Droits : libre de droits, gratuit
Droits réservés à l'éditeur et aux auteurs. © Inria Bordeaux - Sud-Ouest

Auteur(s) : THORPE SIMON
Éditeur(s) : INRIA (Institut national de recherche en informatique et automatique)
10-07-2015

Description : The ability of humans to identify and categorize objects in complex natural scenes has long been thought to be beyond the capacities of artificial vision systems. However, recent progress in Deep Learning and Convolutional Neural Networks has demonstrated that simple feed-forward processing architectures composed of less than 10 layers of neurons can achieve human levels of performance in object recognition tasks. It is interesting to note that such processing architectures have a very similar structure to the primate visual system. Could it be that we are close to understanding how our brains recognize stimuli? I will argue that the main problem with the current state of the art in computer vision is that the learning procedures used are totally unrealistic. Essentially, building such a system requires hundreds of millions of training cycles of supervised learning. By contrast, our own visual systems can learn new stimuli in a few tens of presentations. I will suggest that more biologically realistic learning mechanisms based on spike-based processing and Spike Time Dependent Plasticity (STDP) may be much closer to the way our own visual systems operate, and allow our visual systems to learn about objects in the visual world on the basis of experience.
Mots-clés libres : neurosciences, information visuelle, extraction valeur des images

Classification UNIT : Informatique > Intelligence artificielle : apprentissage, représentation
Systèmes d'information > Fouille de données
Classification : Instruments du savoir : organisations et documents > Informatique
Indice(s) Dewey: Intelligence artificielle, réseaux neuronaux, automates cellulaires, vie artificielle (006.3)


PEDAGOGIQUE

Type pédagogique : cours / présentation

Niveau : doctorat



TECHNIQUE


Type de contenu : image en mouvement
Format : video/x-flv
Taille : 720.69 Mo
Durée d'exécution : 1 heure 25 minutes 45 secondes



RELATIONS


Cette ressource fait partie de :
  • Colloque Interfaces



Entrepôt d'origine : Canal-u.fr
Identifiant : oai:canal-u.fr:23890
Type de ressource : Ressource pédagogique
Exporter au format XML