Les événements du CRIM

Le CRIM organise des événements qui traitent d’une variété de sujets à l’intention de la communauté des technologies de l’information du Québec. Ces rendez-vous, qui favorisent le réseautage et le maillage, abordent des thèmes d’actualité qui vous aideront à prendre les meilleures décisions pour la bonne gestion et le bon développement de vos projets TI.

Edge Implementation of Deep Neural Networks

Edge Implementation of Deep Neural Networks

Présentation en anglais

Réseaux de neurones profonds dans un contexte d'informatique en périphérie

CONFÉRENCIER

Vahid Partovi Nia, Ph. D., Scientifique principal en apprentissage automatique, Huawei


Edge implementation of Deep Neural Networks

SPEAKER

Dr. Vahid Partovi Nia, Principal Machine Learning Scientist, Huawei

ABSTRACT

This talk is divided into two parts. In the first part I focus on network quantization foundation. Low-bit quantization helps to speed up inference, save memory, and ultimately brings the intelligence on edge devices. Motivated from the perceptron, I argue that a quantized network is equivalent to a complex logic. This provides us with a powerful computation vehicle to deploy neural networks only using logical operators. In the second part of the talk I focus on designing compact neural network architectures. I study the exhaustive search foundation as the main powerful neural network design tool. Despite application of the exhaustive search algorithm in many complex problems there is little theory developed in this field. I show some basic theoretical results that give us an insight about convergence conditions of the exhaustive search algorithm.

BIOGRAPHY

Vahid Partovi Nia was born in Mashhad, Iran. He was ranked top <0.1% in 3 national scientific competitions in the field of Statistics namely Mathematical Statistics, Biostatistics, and Actuarial Statistics after completing his bachelor. Vahid continued his M.Sc. in Mathematical Statistics to improve his theoretical knowledge. He successfully held some part-time positions as consultant in university, experiment designer and analyst in automobile industry while studying his master’s degree. He was urban data analyst in municipality before starting his Ph.D.

He obtained his Ph.D. in Mathematics from EPFL Switzerland in 2009 and awarded the Swiss National Science Foundation's prospective researchers competitive fellowship to pursue his research career at McGill and Stanford. In 2011 he joined the department of Mathematics and Industrial Engineering of Ecole Polytechnique de Montreal as a faculty, and moved to Huawei Noah’s Ark Lab in 2018. Since 2019 he is leading the Accelerated Neural Technology (Ant) team as a sub-division of the Computer Vision lab. He is still supervising graduate students in academia, holds Adjunct Professor position at Ecole Polytechnique de Montreal, and is a member of the Canada Excellence Research Chair in Data Science. He also serves in the advisory board of McGill Continuing Studies for their Machine Learning and Artificial Intelligence program. He is an associate member of GERAD optimization center and is a member of STATLab of centre de recherche mathematiques (CRM), a top Canadian research center in Mathematics. Additionally, he holds some ranks in Volleyball and ping pong, is a former Lausanne marathoner, dances latino, plays piano and guitar.


Les séminaires scientifiques du CRIM, gratuits et ouverts à tous, sont donnés par des experts de renommée internationale, des collaborateurs universitaires, le personnel de R-D et les étudiants du CRIM. Au programme, des présentations conviviales sur les dernières avancées scientifiques et technologiques.


Conférence gratuite. Inscription requise.

Le 10 octobre 2019
De 11h à 12h

Lieu CRIM405, avenue Ogilvy,
Bureau 101, Salle 1056
Montréal, H3N 1M3
ContactCRIM514 840-1234

Une présentation de Vahid Partovi Nia, Ph. D., Scientifique principal en apprentissage automatique, Huawei.

Tous les Événements

Liste des événements auxquels le CRIM et ses experts participent. Plusieurs rabais sont offerts à nos membres!

Voir les événements

Vous abonner aux infolettres du CRIM S'inscrire à notre infolettre

Voir nos infolettres récentes

Suivez-nous sur :   LinkedIn Flickr YouTube Medium