BDD4BNN: A BDD-Based Quantitative Analysis Framework for Binarized Neural Networks
2021
会议录名称PROCEEDINGS OF THE 33RD INTERNATIONAL CONFERENCE ON COMPUTER AIDED VERIFICATION, CAV 2021 (IF:0.402[JCR-2005],0.000[5-Year])
ISSN0302-9743
卷号12759
页码175-200
发表状态已发表
DOI10.1007/978-3-030-81685-8_8
摘要

Verifying and explaining the behavior of neural networks is becoming increasingly important, especially when they are deployed in safety-critical applications. In this paper, we study verification and interpretability problems for Binarized Neural Networks (BNNs), the 1-bit quantization of general real-numbered neural networks. Our approach is to encode BNNs into Binary Decision Diagrams (BDDs), which is done by exploiting the internal structure of the BNNs. In particular, we translate the input-output relation of blocks in BNNs to cardinality constraints which are in turn encoded by BDDs. Based on the encoding, we develop a quantitative framework for BNNs where precise and comprehensive analysis of BNNs can be performed. We demonstrate the application of our framework by providing quantitative robustness analysis and interpretability for BNNs. We implement a prototype tool BDD4BNN and carry out extensive experiments, confirming the effectiveness and efficiency of our approach.

收录类别EI ; CPCI ; CPCI-S
语种英语
WOS研究方向Computer Science
WOS类目Computer Science, Hardware & Architecture ; Computer Science, Software Engineering ; Computer Science, Theory & Methods
WOS记录号WOS:000698732400008
出版者SPRINGER INTERNATIONAL PUBLISHING AG
原始文献类型Proceedings Paper
引用统计
正在获取...
文献类型会议论文
条目标识符https://kms.shanghaitech.edu.cn/handle/2MSLDSTB/128429
专题信息科学与技术学院_博士生
信息科学与技术学院_PI研究组_宋富组
通讯作者Song, Fu
作者单位
1.ShanghaiTech Univ, Shanghai, Peoples R China;
2.Shanghai Engn Res Ctr Intelligent Vis & Imaging, Shanghai, Peoples R China;
3.Birkbeck Univ London, London, England
第一作者单位上海科技大学
通讯作者单位上海科技大学
第一作者的第一单位上海科技大学
推荐引用方式
GB/T 7714
Zhang, Yedi,Zhao, Zhe,Chen, Guangke,et al. BDD4BNN: A BDD-Based Quantitative Analysis Framework for Binarized Neural Networks[C]:SPRINGER INTERNATIONAL PUBLISHING AG,2021:175-200.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
个性服务
查看访问统计
谷歌学术
谷歌学术中相似的文章
[Zhang, Yedi]的文章
[Zhao, Zhe]的文章
[Chen, Guangke]的文章
百度学术
百度学术中相似的文章
[Zhang, Yedi]的文章
[Zhao, Zhe]的文章
[Chen, Guangke]的文章
必应学术
必应学术中相似的文章
[Zhang, Yedi]的文章
[Zhao, Zhe]的文章
[Chen, Guangke]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。