Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Denton, Remi, Zaremba, Wojciech, Bruna, Joan, LeCun, Yann, Fergus, Rob
Natura: Preprint
Pubblicazione: 2014
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909135148154880
author Denton, Remi
Zaremba, Wojciech
Bruna, Joan
LeCun, Yann
Fergus, Rob
author_facet Denton, Remi
Zaremba, Wojciech
Bruna, Joan
LeCun, Yann
Fergus, Rob
contents We present techniques for speeding up the test-time evaluation of large convolutional networks, designed for object recognition tasks. These models deliver impressive accuracy but each image evaluation requires millions of floating point operations, making their deployment on smartphones and Internet-scale clusters problematic. The computation is dominated by the convolution operations in the lower layers of the model. We exploit the linear structure present within the convolutional filters to derive approximations that significantly reduce the required computation. Using large state-of-the-art models, we demonstrate we demonstrate speedups of convolutional layers on both CPU and GPU by a factor of 2x, while keeping the accuracy within 1% of the original model.
format Preprint
id arxiv_https___arxiv_org_abs_1404_0736
institution arXiv
publishDate 2014
record_format arxiv
spellingShingle Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation
Denton, Remi
Zaremba, Wojciech
Bruna, Joan
LeCun, Yann
Fergus, Rob
Computer Vision and Pattern Recognition
Machine Learning
We present techniques for speeding up the test-time evaluation of large convolutional networks, designed for object recognition tasks. These models deliver impressive accuracy but each image evaluation requires millions of floating point operations, making their deployment on smartphones and Internet-scale clusters problematic. The computation is dominated by the convolution operations in the lower layers of the model. We exploit the linear structure present within the convolutional filters to derive approximations that significantly reduce the required computation. Using large state-of-the-art models, we demonstrate we demonstrate speedups of convolutional layers on both CPU and GPU by a factor of 2x, while keeping the accuracy within 1% of the original model.
title Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/1404.0736