针对心电图分类模型的平滑攻击算法

doi:10.13190/j.jbupt.2022-031

北京邮电大学学报 ›› 2022, Vol. 45 ›› Issue (4): 44-50.doi: 10.13190/j.jbupt.2022-031

针对心电图分类模型的平滑攻击算法

刘金桐¹, 杨国兴¹, 刘晓鸿², 王光宇¹

1. 北京邮电大学信息与通信工程学院, 北京 100876;
2. 清华大学计算机科学与技术系, 北京 100084

收稿日期:2022-02-06 出版日期:2022-08-28 发布日期:2022-06-26
通讯作者: 王光宇(1990—),女,研究员,博士生导师,邮箱:guangyu.wang@bupt.edu.cn。 E-mail:guangyu.wang@bupt.edu.cn
作者简介:刘金桐(1999—),男,硕士生。
基金资助:
国家重点研发计划项目（2019YFB1404804）；国家自然科学基金项目（61906105）

Smoothing Attack Algorithm Based on Electrocardiogram Classification

LIU Jintong¹, YANG Guoxing¹, LIU Xiaohong², WANG Guangyu¹

1. School of Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China;
2. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China

Received:2022-02-06 Online:2022-08-28 Published:2022-06-26

摘要/Abstract

摘要： 在心电图分类领域中，传统攻击算法生成的对抗样本存在生理上不可解释的方波且生成效率低下，为此，提出了一种补丁平滑攻击（PatchSAP）算法。针对卷积神经网络、长短记忆网络和基于注意力机制的长短期记忆网络3种常见心电图分类模型开展对抗攻击，比较了心电分类模型的"脆弱"程度并分析了模型超参数。实验结果表明，PatchSAP算法的攻击效率与传统攻击算法的攻击效率相比具有明显的优势，生成的对抗样本能很好地维持真实性，卷积核和约束范围等超参数对对抗样本的有效性和真实性有较大影响。

关键词: 单导联心电图, 深度学习, 对抗攻击, 平滑攻击

Abstract: In the field of electrocardiogram classification, the adversarial samples generated by the traditional projected gradient algorithm with low generation efficiency have square waves that cannot be explained physiologically, and thus, a patch-based smooth attack perturbations (PatchSAP) algorithm is proposed. By conducting adversarial attacks against three common electrocardiogram classification models, convolutional neural network, long-short-term memory network, and attention-based long-short-term memory network, we compare the "vulnerability" of the electrocardiogram classification models, and analyze the hyperparameter to obtain the difference between validity and authenticity of adversarial examples. The experimental results show that the PatchSAP algorithm has obvious advantages in attack efficiency, and the generated adversarial samples maintain the sample authenticity well. Hyperparameters such as convolution kernel and constraint range have a great impact on the effectiveness and authenticity of adversarial examples.

Key words: single lead electrocardiogram, deep learning, adversarial attack, smoothing attack

中图分类号:

TP183

刘金桐, 杨国兴, 刘晓鸿, 王光宇. 针对心电图分类模型的平滑攻击算法[J]. 北京邮电大学学报, 2022, 45(4): 44-50.

LIU Jintong, YANG Guoxing, LIU Xiaohong, WANG Guangyu. Smoothing Attack Algorithm Based on Electrocardiogram Classification[J]. Journal of Beijing University of Posts and Telecommunications, 2022, 45(4): 44-50.

参考文献

[1] KAPLAN B S, UYSAL A K, SORA G E, et al. A survey on ECG analysis[J]. Biomedical Signal Processing and Control, 2018, 43:216-235.
[2] HANNUN A Y, RAJPURKAR P, HAGHPANAHI M, et al. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network[J]. Nature Medicine, 2019, 25(1):65-69.
[3] MOOSAVI-DEZFOOLI S M, FAWZI A, FAWZI O, et al. Universal adversarial perturbations[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE Press, 2017:86-94.
[4] WANG K, WANG G Y, CHEN N, et al. How robust is your automatic diagnosis model?[C]//2019 IEEE International Conference on Bioinformatics and Biomedicine. Piscataway, NJ:IEEE Press, 2019:877-884.
[5] HAN X T, HU Y X, FOSCHINI L, et al. Deep learning models for electrocardiograms are susceptible to adversarial attack[J]. Nature Medicine, 2020, 26(3):360-363.
[6] KIRANYAZ S, INCE T, GABBOUJ M. Real-time patient-specific ECG classification by 1-D convolutional neural networks[J]. IEEE Transactions on Biomedical Engineering, 2016, 63(3):664-675.
[7] RANA A, KIM K K. ECG heartbeat classification using a single layer LSTM model[C]//2019 International SoC Design Conference. Piscataway, NJ:IEEE Press, 2019:267-268.
[8] LIU F, ZHOU X S, WANG T B, et al. An attention-based hybrid LSTM-CNN model for arrhythmias classification[C]//IJCNN 2019:International Joint Conference on Neural Networks. Piscataway, NJ:IEEE Press, 2019:1-8.
[9] KARIMIAN N, WOODARD D L, FORTE D. On the vulnerability of ECG verification to online presentation attacks[C]//2017 IEEE International Joint Conference on Biometrics. Piscataway, NJ:IEEE Press, 2017:143-151.
[10] MA L H, LIANG L. Enhance CNN robustness against noises for classification of 12-lead ECG with variable length[C]//2020 19th IEEE International Conference on Machine Learning and Applications. Piscataway, NJ:IEEE Press, 2020:839-846.
[11] 徐晨阳, 李梦昕, 杨娟. 超声图像神经分割方法研究[J]. 北京邮电大学学报, 2018, 41(1):115-120. XU C Y, LI M X, YANG J. Neural segmentation method of ultrasound image[J]. Journal of Beijing University of Posts and Telecommunications, 2018, 41(1):115-120.

[1]	刘阳, 滕颖蕾, 牛涛, 郅佳琳. 基于深度强化学习的滤波器剪枝方案[J]. 北京邮电大学学报, 2023, 46(3): 31-36.
[2]	靳梦凡, 黄智濒, 储志强. 复杂环境下多模态指导的点云补全方法[J]. 北京邮电大学学报, 2023, 46(3): 103-108.
[3]	荣震宇刘建毅. 基于Transformer和MLP的眼底血管分割算法[J]. 北京邮电大学学报, 2023, 46(1): 26-31.
[4]	黄柳婷, 刘可欣, 牛凯, 常春, 贺志强. 基于深度学习的哮喘患者CT影像黏液栓自动识别[J]. 北京邮电大学学报, 2022, 45(4): 58-63.
[5]	柳长源, 虎浩媛, 毕晓君. 双线性融合网络的驾驶员分心行为识别[J]. 北京邮电大学学报, 2022, 45(2): 79-84.
[6]	赵海英, 朱会, 侯小刚. 基于改进EMA单元的传统服饰图像语义分割[J]. 北京邮电大学学报, 2022, 45(1): 69-74.
[7]	张滨宇, 赵衍运, 杜昀昊, 万俊峰, 佟知航. 一种基于深度学习的PCB图像字符检测方法[J]. 北京邮电大学学报, 2022, 45(1): 108-114.
[8]	贾珺, 冯春燕, 夏海轮, 张天魁, 李成钢. 基于样本均衡与特征交互的通信网络故障预测方法[J]. 北京邮电大学学报, 2021, 44(6): 59-66.
[9]	普运伟, 郭江, 刘涛涛, 吴海潇. 基于多学习单元卷积神经网络的雷达辐射源信号识别[J]. 北京邮电大学学报, 2021, 44(6): 74-82.
[10]	王艺霏, 莫爽, 吴文睿, 范少华, 肖丁. 基于内外卷积网络的网络入侵检测[J]. 北京邮电大学学报, 2021, 44(5): 94-100.
[11]	高慧, 张继威, 来扬, 王文东. 深度学习的人体图像半自动标注系统[J]. 北京邮电大学学报, 2021, 44(1): 104-109.
[12]	蒲悦逸, 王文涵, 朱强, 陈朋朋. 基于CNN-ResNet-LSTM模型的城市短时交通流量预测算法[J]. 北京邮电大学学报, 2020, 43(5): 9-14.
[13]	马向亮, 李冰, 杨丹, 黄克振, 段晓毅. 基于深度学习的类SM4算法S盒逆向分析[J]. 北京邮电大学学报, 2020, 43(5): 118-124.
[14]	张天麒, 康波, 孟祥飞, 刘奕琳, 周颖. 基于U-Net的颅内出血识别算法[J]. 北京邮电大学学报, 2020, 43(3): 92-98.
[15]	马素刚, 赵祥模, 侯志强, 王忠民, 孙韩林. 一种基于ResNet网络特征的视觉目标跟踪算法[J]. 北京邮电大学学报, 2020, 43(2): 129-134.

针对心电图分类模型的平滑攻击算法

Smoothing Attack Algorithm Based on Electrocardiogram Classification

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价