轻量化深度卷积神经网络设计研究进展  

Research Progress on Designing Lightweight Deep Convolutional Neural Networks

作　　者：周志飞 李华[1,2] 冯毅雄 陆见光 钱松荣 李少波[1] ZHOU Zhifei;LI Hua;FENG Yixiong;LU Jianguang;QIAN Songrong;LI Shaobo(State Key Laboratory of Public Big Data,Guizhou University,Guiyang 550025,China;Department of Mechanical Engineering,Tsinghua University,Beijing 100084,China)

机构地区：[1]贵州大学省部共建公共大数据国家重点实验室,贵阳550025 [2]清华大学机械工程系,北京100084

出　　处：《计算机工程与应用》2024年第22期1-17,共17页Computer Engineering and Applications

基　　金：国家自然科学基金青年基金项目(52205092);中国博士后面上项目(2023M731939);贵州大学自然科学专项(特岗)科研项目([2021]27);贵州大学培育项目([2020]25)。

摘　　要：轻量化设计是解决深度卷积神经网络(deep convolutional neural network,DCNN)对设备性能和硬件资源依赖性的流行范式,轻量化的目的是在不牺牲网络性能的前提下,提高计算速度和减少内存占用。综述了DCNN的轻量化设计方法,着重回顾了近年来DCNN的研究进展,包括体系设计和模型压缩两大轻量化策略,深入比较了这两类方法的创新性、优势与局限性,并探讨了支撑轻量化模型的底层框架。此外,对轻量化网络已经成功应用的场景进行了描述,并对DCNN轻量化的未来发展趋势进行了预测,旨在为深度卷积神经网络的轻量化研究提供有益的见解和参考。Lightweight design is a popular paradigm to address the dependence of deep convolutional neural network(DCNN)on device performance and hardware resources,and the purpose of lightweighting is to increase the computational speed and reduce the memory footprint without sacrificing the network performance.An overview of lightweight design approaches for DCNNs is presented,focusing on a review of the research progress in recent years,including two major lightweighting strategies,namely,system design and model compression,as well as an in-depth comparison of the innova-tiveness,strengths and limitations of these two types of approaches,and an exploration of the underlying framework that supports the lightweighting model.In addition,scenarios in which lightweight networks have been successfully applied are described,and predictions are made for the future development trend of DCNN lightweighting,aiming to provide use-ful insights and references for the research on lightweight deep convolutional neural networks.

关 键 词：深度卷积神经网络 轻量化 体系设计 模型压缩 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]