arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2003.09372cs.LGstat.ML
收藏

一个神经元愚弄所有人

One Neuron to Fool Them All

University of Virginia(弗吉尼亚大学)

Anshuman Suri, David Evans

摘要:

尽管对对抗性例子进行了大量研究,但模型易感性的根本原因还没有得到很好的理解。我们提出了一个概念,而不是看攻击特定的鲁棒性,评估单个神经元的敏感性方面如何强大的模型的输出是直接扰动的神经元的输出。从这个角度分析模型,可以发现标准模型和经过逆向训练的鲁棒模型的独特特征,并导致一些奇怪的结果。在我们对CIFAR-10和ImageNet的实验中,我们发现使用仅针对单个敏感神经元的损失函数的攻击几乎与针对完整模型的攻击一样有效。我们分析了这些敏感神经元的属性,提出了一个正则化项,可以帮助模型实现对各种不同扰动约束的鲁棒性,同时保持对自然数据分布的准确性。我们所有实验的代码都可以在https://github.com/iamgroot42/sauron上找到。

英文摘要:

Despite vast research in adversarial examples, the root causes of model susceptibility are not well understood. Instead of looking at attack-specific robustness, we propose a notion that evaluates the sensitivity of individual neurons in terms of how robust the model's output is to direct perturbations of that neuron's output. Analyzing models from this perspective reveals distinctive characteristics of standard as well as adversarially-trained robust models, and leads to several curious results. In our experiments on CIFAR-10 and ImageNet, we find that attacks using a loss function that targets just a single sensitive neuron find adversarial examples nearly as effectively as ones that target the full model. We analyze the properties of these sensitive neurons to propose a regularization term that can help a model achieve robustness to a variety of different perturbation constraints while maintaining accuracy on natural data distributions. Code for all our experiments is available at https://github.com/iamgroot42/sauron .

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。

↑