Review of Lightweight Deep Convolutional Neural Networks

Lightweight deep convolutional neural networks (LDCNNs) are vital components of mobile intelligence, particularly in mobile vision. Although various heavy networks with increasingly deeper and wider have continuously broken accuracy records since 2012, with the spring of terminals and mobile devices...

Full description

Saved in:
Bibliographic Details
Published in:Archives of computational methods in engineering 2024-05, Vol.31 (4), p.1915-1937
Main Authors: Chen, Fanghui, Li, Shouliang, Han, Jiale, Ren, Fengyuan, Yang, Zhen
Format: Article
Language:English
Subjects:
Artificial neural networks
Engineering
Intelligence
Lightweight
Mathematical and Computational Engineering
Neural networks
Review Article
Weight reduction
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Lightweight deep convolutional neural networks (LDCNNs) are vital components of mobile intelligence, particularly in mobile vision. Although various heavy networks with increasingly deeper and wider have continuously broken accuracy records since 2012, with the spring of terminals and mobile devices, neural networks that can match them have become a core role in practical applications. In this review, we focus on several representative lightweight Deep Convolutional Neural Networks (DCNN) technologies that hold significant potential for advancing the field. More than 190 references screened out in terms of architecture design and model compression, in which over 50 representative ones are emphasized from the perspectives of methods, performance, advantages, and drawbacks, as well as underlying framework support and benchmark datasets. With a comprehensive analysis, we put forward some existing problems and offer prospects of lightweight DCNN for future development.
ISSN:1134-3060
1886-1784
DOI:10.1007/s11831-023-10032-z