Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Research Tools and Instruments - Category 1 (<$150,000)

Université de Montréal

$111,762.00

GPUs for unsupervised learning of representations

EF 2009-10

Résumé
New artificial intelligence algorithms called deep architectures were recently proposed in the machine learning community that have the potential to extract meaningful representations from vast amounts of data such as the images in the web and music collections. A meaningful representation for an image is one that captures what humans care about, e.g., what are the objects in it, but is not too sensitive to changes due to factors such as position, illumination, scale, etc. Similarly, meaningful representations of music separate factors such as timbre from the melodic content and other factors involved. The proposed equipment will significantly boost the research program of the U. Montreal machine learning laboratory (www.iro.umontreal.ca/~lisa) in the area of unsupervised learning of representations using these new algorithms, in particular for images and music. Whereas current computers are too slow for large-scale implementation of these algorithms, even with large computer clusters, the proposed equipment promises 10-fold to 100-fold speed-ups in training time thanks to the advent of Graphics Processing Units (GPUs) exploited as general-purpose parallel computing devices. A recent paper from Stanford university on the potential of GPUs in this domain states that they have the "potential to revolutionize the applicability of deep unsupervised learning methods" and most labs investigating these techniques are now developing their algorithms with GPUs.
Programme
Research Tools and Instruments - Category 1 (<$150,000)
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
389508-2010
Name nom
Bengio, Yoshua
Department département
Informatique et recherche opérationnelle
Organization id
63
Institution établissement
Université de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2009
Competition year année de concours
2010
Award amount
111762
Program id
EQPEQ
Program nane en
Research Tools and Instruments - Category 1 (<$150,000)
Program name fr
Outils et instruments de recherche - catégorie 1 (<150 000$)
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
800
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information and communication services
Area of application fr
Services d'information et de télécommunications
Research subject code
2805
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Learning and inference theories
Research subject fr
Théories de l'apprentissage et l'inférence
Partie
1
Nb partie
1
Application title
GPUs for unsupervised learning of representations
Keyword
NA
Application summary
New artificial intelligence algorithms called deep architectures were recently proposed in the machine learning community that have the potential to extract meaningful representations from vast amounts of data such as the images in the web and music collections. A meaningful representation for an image is one that captures what humans care about, e.g., what are the objects in it, but is not too sensitive to changes due to factors such as position, illumination, scale, etc. Similarly, meaningful representations of music separate factors such as timbre from the melodic content and other factors involved. The proposed equipment will significantly boost the research program of the U. Montreal machine learning laboratory (www.iro.umontreal.ca/~lisa) in the area of unsupervised learning of representations using these new algorithms, in particular for images and music. Whereas current computers are too slow for large-scale implementation of these algorithms, even with large computer clusters, the proposed equipment promises 10-fold to 100-fold speed-ups in training time thanks to the advent of Graphics Processing Units (GPUs) exploited as general-purpose parallel computing devices. A recent paper from Stanford university on the potential of GPUs in this domain states that they have the "potential to revolutionize the applicability of deep unsupervised learning methods" and most labs investigating these techniques are now developing their algorithms with GPUs.