利用非负权重约束防御恶意软件分类网络对抗对抗性扰动
Defending Malware Classification Networks Against Adversarial Perturbations with Non-Negative Weight Restrictions
摘要：
越来越多的文献表明，深度神经网络容易受到对抗性输入修改的影响。最近，这项工作已经从图像分类扩展到布尔特征上的恶意软件分类。在本文中，我们提出了几种在这个特定领域训练受限网络的新方法，这些方法在防止对抗性扰动方面非常有效。我们从一个完全抗对抗的神经网络开始，它具有严格的非负权重限制，相当于学习单调布尔函数，然后尝试放松约束以提高分类器的准确性。
英文摘要：
There is a growing body of literature showing that deep neural networks are vulnerable to adversarial input modification. Recently this work has been extended from image classification to malware classification over boolean features. In this paper we present several new methods for training restricted networks in this specific domain that are highly effective at preventing adversarial perturbations. We start with a fully adversarially resistant neural network that has hard non-negative weight restrictions and is equivalent to learning a monotonic boolean function and then attempt to relax the constraints to improve classifier accuracy.
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。
