CASIA OpenIR  > 图像与视频分析团队
Training Binary Weight Networks via Semi-Binary Decomposition
Hu, Qinghao1,2; Li, Gang1,2; Wang, Peisong1,2; Zhang, Yifan1,2; Cheng, Jian1,2,3
2018-09
Conference NameEuropean Conference on Computer Vision (ECCV).2018
Conference Date2018-9
Conference Place德国慕尼黑
Abstract

Recently binary weight networks have attracted lots of attentions due to their high computational efficiency and small parameter size. Yet they still suffer from large accuracy drops because of their limited representation capacity. 
In this paper, we propose a novel semi-binary decomposition method which decomposes a matrix into two binary matrices and a diagonal matrix. Since the matrix product of binary matrices has more numerical values than binary matrix, 
the proposed semi-binary decomposition has more representation capacity. 
Besides, we propose an alternating optimization method to solve the semi-binary decomposition problem while keeping binary constraints.
Extensive experiments on AlexNet, ResNet-18, and ResNet-50 demonstrate that our method outperforms state-of-the-art methods by a large margin (5 percentage higher in top1 accuracy). We also implement binary weight AlexNet on FPGA platform, which shows that our proposed method can achieve $\sim 9\times$ speed-ups while reducing the consumption of on-chip memory and dedicated multipliers significantly.

Indexed ByEI
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/23705
Collection图像与视频分析团队
Corresponding AuthorCheng, Jian
Affiliation1.Institute of Automation, Chinese Academy of Sciences, Beijing, China
2.University of Chinese Academy of Sciences, Beijing, China
3.Center for Excellence in Brain Science and Intelligence Technology, Beijing, China
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Corresponding Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Hu, Qinghao,Li, Gang,Wang, Peisong,et al. Training Binary Weight Networks via Semi-Binary Decomposition[C],2018.
Files in This Item: Download All
File Name/Size DocType Version Access License
0825.pdf(575KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Hu, Qinghao]'s Articles
[Li, Gang]'s Articles
[Wang, Peisong]'s Articles
Baidu academic
Similar articles in Baidu academic
[Hu, Qinghao]'s Articles
[Li, Gang]'s Articles
[Wang, Peisong]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Hu, Qinghao]'s Articles
[Li, Gang]'s Articles
[Wang, Peisong]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 0825.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.