CP-decomposition with Tensor Power Method for Convolutional Neural Networks compression

Convolutional Neural Networks (CNNs) has shown a great success in many areas including complex image classification tasks. However, they need a lot of memory and computational cost, which hinders them from running in relatively low-end smart devices such as smart phones. We propose a CNN compression...

Full description

Saved in:
Bibliographic Details
Published inInternational Conference on Big Data and Smart Computing pp. 115 - 118
Main Authors Astrid, Marcella, Seung-Ik Lee
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.02.2017
Subjects
Convolution
Kernel
Matrix decomposition
Neural networks
Tensile stress
Tuning
Online AccessGet full text
ISSN2375-9356
DOI10.1109/BIGCOMP.2017.7881725

Cover

More Information
Summary:Convolutional Neural Networks (CNNs) has shown a great success in many areas including complex image classification tasks. However, they need a lot of memory and computational cost, which hinders them from running in relatively low-end smart devices such as smart phones. We propose a CNN compression method based on CP-decomposition and Tensor Power Method. We also propose an iterative fine tuning, with which we fine-tune the whole network after decomposing each layer, but before decomposing the next layer. Significant reduction in memory and computation cost is achieved compared to state-of-the-art previous work with no more accuracy loss.
ISSN:2375-9356
DOI:10.1109/BIGCOMP.2017.7881725