期刊
JOURNAL OF NEURAL ENGINEERING
卷 18, 期 4, 页码 -出版社
IOP Publishing Ltd
DOI: 10.1088/1741-2552/ac0f4c
关键词
brain-computer interface; convolutional neural network; electroencephalogram; universal adversarial perturbation
资金
- Technology Innovation Project of Hubei Province of China [2019AEA171]
- Zhejiang Lab [2021KE0AB04]
- Hubei Province Funds for Distinguished Young Scholars [2020CFA050]
- National Natural Science Foundation of China [61873321, U1913207]
- National Social Science Foundation of China [19ZDA104, 20AZD089]
- International Science and Technology Cooperation Program of China [2017YFE0128300]
The study introduces a novel total loss minimization approach to generate universal adversarial perturbations for EEG-based BCIs. Experimental results demonstrate the effectiveness of this method on three popular CNN classifiers for both target and non-target attacks, while also confirming the transferability of UAPs across different systems.
Objective. Multiple convolutional neural network (CNN) classifiers have been proposed for electroencephalogram (EEG) based brain-computer interfaces (BCIs). However, CNN models have been found vulnerable to universal adversarial perturbations (UAPs), which are small and example-independent, yet powerful enough to degrade the performance of a CNN model, when added to a benign example. Approach. This paper proposes a novel total loss minimization (TLM) approach to generate UAPs for EEG-based BCIs. Main results. Experimental results demonstrated the effectiveness of TLM on three popular CNN classifiers for both target and non-target attacks. We also verified the transferability of UAPs in EEG-based BCI systems. Significance. To our knowledge, this is the first study on UAPs of CNN classifiers in EEG-based BCIs. UAPs are easy to construct, and can attack BCIs in real-time, exposing a potentially critical security concern of BCIs.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据