Spiking neural networks (SNNs) have attracted much attention for their high energy efficiency and for recent advances in their classification performance. However, unlike traditional deep learning approaches, the analysis and study of the robustness of SNNs to adversarial examples remain relatively underdeveloped. In this work we focus on advancing the adversarial attack side of SNNs and make three major contributions. First, we show that successful white-box adversarial attacks on SNNs are highly dependent on the underlying surrogate gradient technique. Second, using the best surrogate gradient technique, we analyze the transferability of adversarial attacks on SNNs and other state-of-the-art architectures like Vision Transformers (ViTs) and Big Transfer Convolutional Neural Networks (CNNs). We demonstrate that SNNs are not often deceived by adversarial examples generated by Vision Transformers and certain types of CNNs. Third, due to the lack of an ubiquitous white-box attack that is effective across both the SNN and CNN/ViT domains, we develop a new white-box attack, the Auto Self-Attention Gradient Attack (Auto SAGA). Our novel attack generates adversarial examples capable of fooling both SNN models and non-SNN models simultaneously. Auto SAGA is as much as $87.9\%$ more effective on SNN/ViT model ensembles than conventional white-box attacks like PGD. Our experiments and analyses are broad and rigorous covering three datasets (CIFAR-10, CIFAR-100 and ImageNet), five different white-box attacks and nineteen different classifier models (seven for each CIFAR dataset and five different models for ImageNet).
翻译:Spik 神经网络(SNNS)因其高能效和最近分类绩效的进步而引起人们的极大关注。然而,与传统的深层次学习方法不同,分析和研究SNNS对对抗性实例的稳健性仍然相对不足。在这项工作中,我们的重点是推进SNNS的对抗性攻击方,并做出三大贡献。首先,我们表明,对SNNS的成功白箱对抗性攻击高度依赖基础替代梯度技术。第二,使用最佳代金梯度技术,我们分析对SNNS和其他最先进的结构的对抗性攻击的可转移性,如Vision Trangers(Vivision Trangers)和大型转移性神经网络网络网络(Bigal-Nations)的强势性攻击性攻击性攻击性攻击性,以及S-NFRFR 和CNNV/VT两个领域,我们开发新的白箱攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性, 自动自控性自控性FRRFRER AL 5型攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性数字性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性数字性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性模型(SMASMASMASMAS型模型,甚为SMAS-SAS-SASMAS-SAS-SAMA性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性飞机性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击性攻击