PARS: Proxy-Based Automatic Rank Selection for Neural Network Compression via Low-Rank Weight Approximation

PARS: Proxy-Based Automatic Rank Selection for Neural Network Compression via Low-Rank Weight Approximation

Low-rank matrix/tensor decompositions are promising methods for reducing the inference time, computation, and memory consumption of deep neural networks (DNNs). This group of methods decomposes the pre-trained neural network weights through low-rank matrix/tensor decomposition and replaces the origi...

Full description

Bibliographic Details
Main Authors: Konstantin Sobolev, Dmitry Ermilov, Anh-Huy Phan, Andrzej Cichocki
Format: Article
Language:English
Published: MDPI AG 2022-10-01
Series:Mathematics
Subjects:
convolutional neural network acceleration
deep learning
low-rank tensor decomposition
rank selection
Online Access:https://www.mdpi.com/2227-7390/10/20/3801

Internet

https://www.mdpi.com/2227-7390/10/20/3801

Similar Items