arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.06032cs.CVcs.CRcs.LG
收藏

基于概率雅可比的显著图攻击

Probabilistic Jacobian-based Saliency Maps Attacks

CentraleSupélec(中央理工-高等电力学院);IRT SystemX(SystemX技术研究院)

Théo Combey, António Loison, Maxime Faucher, Hatem Hajri

摘要:

已知神经网络分类器(NNC)容易受到输入的恶意对抗性扰动,包括那些修改称为稀疏或$L_0$攻击的输入特征的一小部分的扰动。有效且快速的$L_0$攻击,例如广泛使用的基于雅可比的显著图攻击(JSMA),不仅可以欺骗NNC,还可以提高其鲁棒性。在本文中,我们表明,惩罚显着图的JSMA的输出概率和输入功能的NNC允许获得更强大的攻击算法,更好地考虑到每个输入的特性。这导致我们引入JSMA的改进版本,命名为加权JSMA(WJSMA)和泰勒JSMA(TJSMA),并通过在三个不同数据集(MNIST,CIFAR-10和GTSRB)上的各种白盒和黑盒实验证明,它们比JSMA的原始目标和非目标版本更快,更有效。实验还表明,在某些情况下,与Carlini-Wagner(CW)$L_0$攻击相比,我们的攻击具有非常有竞争力的结果,同时像JSMA一样,速度明显更快(WJSMA和TJSMA在CIFAR-10上比CW $L_0$快50倍以上)。因此,我们的新攻击为$L_0$在数据集上进行实时对抗测试提供了JSMA和CW之间的良好权衡,例如之前引用的数据集。代码可通过链接https://github.com/probabilistic-jsmas/probabilistic-jsmas公开获取。

英文摘要:

Neural network classifiers (NNCs) are known to be vulnerable to malicious adversarial perturbations of inputs including those modifying a small fraction of the input features named sparse or $L_0$ attacks. Effective and fast $L_0$ attacks, such as the widely used Jacobian-based Saliency Map Attack (JSMA) are practical to fool NNCs but also to improve their robustness. In this paper, we show that penalising saliency maps of JSMA by the output probabilities and the input features of the NNC allows to obtain more powerful attack algorithms that better take into account each input's characteristics. This leads us to introduce improved versions of JSMA, named Weighted JSMA (WJSMA) and Taylor JSMA (TJSMA), and demonstrate through a variety of white-box and black-box experiments on three different datasets (MNIST, CIFAR-10 and GTSRB), that they are both significantly faster and more efficient than the original targeted and non-targeted versions of JSMA. Experiments also demonstrate, in some cases, very competitive results of our attacks in comparison with the Carlini-Wagner (CW) $L_0$ attack, while remaining, like JSMA, significantly faster (WJSMA and TJSMA are more than 50 times faster than CW $L_0$ on CIFAR-10). Therefore, our new attacks provide good trade-offs between JSMA and CW for $L_0$ real-time adversarial testing on datasets such as the ones previously cited. Codes are publicly available through the link https://github.com/probabilistic-jsmas/probabilistic-jsmas.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑