《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (9): 2686-2691.DOI: 10.11772/j.issn.1001-9081.2022091392

• 2022第10届CCF大数据学术会议 • 上一篇    下一篇

基于混合机制的深度神经网络压缩算法

赵旭剑(), 李杭霖   

  1. 西南科技大学 计算机科学与技术学院,四川 绵阳 621010
  • 收稿日期:2022-09-19 修回日期:2022-10-24 接受日期:2022-10-27 发布日期:2023-01-10 出版日期:2023-09-10
  • 通讯作者: 赵旭剑
  • 作者简介:李杭霖(2000—),女,四川成都人,硕士研究生,主要研究方向:深度学习、神经网络压缩。
  • 基金资助:
    教育部人文社会科学基金资助项目(17YJCZH260);四川省科学技术厅重点研发项目(2020YFS0057);赛尔网络下一代互联网技术创新项目(NGII20180403)

Deep neural network compression algorithm based on hybrid mechanism

Xujian ZHAO(), Hanglin LI   

  1. School of Computer Science and Technology,Southwest University of Science and Technology,Mianyang Sichuan 621010,China
  • Received:2022-09-19 Revised:2022-10-24 Accepted:2022-10-27 Online:2023-01-10 Published:2023-09-10
  • Contact: Xujian ZHAO
  • About author:LI Hanglin, born in 2000, M. S. candidate. Her research interests include deep learning, neural network compression.
  • Supported by:
    Humanities and Social Science Foundation of Ministry of Education(17YJCZH260);Key Research and Development Project of Science and Technology Department of Sichuan Province(2020YFS0057);CERNET Innovation Project(NGII20180403)

摘要:

近年来人工智能(AI)应用飞速发展,嵌入式设备与移动设备等有限资源设备对深度神经网络(DNN)的需求急剧增加。如何在不影响DNN效果的基础上对神经网络进行压缩具有极大理论与现实意义,也是当下深度学习的热门研究话题。首先,针对DNN因模型大、计算量大而难以移植至移动设备等有限资源设备的问题,深入分析已有DNN压缩算法在内存占用、运行速度及压缩效果等方面的实验性能,从而挖掘DNN压缩算法的影响要素;然后,设计学生网络和教师网络组成的知识迁移结构,融合知识蒸馏、结构设计、网络剪枝和参数量化机制,提出基于混合机制的DNN优化压缩算法。在mini-ImageNet数据集上以AlexNet为Benchmark,进行实验比较与分析。实验结果表明,所提算法在压缩结果的准确率降低6.3%的情况下,使压缩后的AlexNet的容量减小98.5%,验证了所提算法的有效性。

关键词: 深度神经网络, 网络压缩, 网络剪枝, 知识蒸馏, 参数量化

Abstract:

With the rapid development of Artificial Intelligence (AI) in recent years, the demand for Deep Neural Network (DNN) from devices with limited resources such as embedded devices and mobile devices has increased sharply. The problem of how to compress neural networks without affecting the effect of DNNs has great theoretical and practical significance, and is a hot research topic in deep learning now. Firstly, aiming at the problem that DNN is difficult to be ported to resource-limited devices such as mobile devices due to their large models and large computational cost, the experimental performance of existing DNN compression algorithms in terms of memory usage, running speed, and compression effect was deeply analyzed, so that the influence factors of the DNN compression algorithm were explored. Then, the knowledge transfer structure composed of student network and teacher network was designed, the knowledge distillation, structural design, network pruning, and parameter quantization mechanisms were fused together, and a DNN optimization and compression model based on hybrid mechanism was proposed. Experimental comparison and analysis were conducted on mini-ImageNet dataset using AlexNet as the Benchmark. Experimental results show that the capacity of compressed AlexNet is reduced by 98.5% with 6.3% loss of accuracy, which verify the effectiveness of the proposed algorithm.

Key words: Deep Neural Network (DNN), network compression, network pruning, knowledge distillation, parameter quantization

中图分类号: