对抗训练需要使用整个训练数据集吗？
Do we need entire training data for adversarial training?
AI总结：
针对对抗训练计算成本高的问题，提出通过筛选易受对抗攻击的样本子集进行训练，在保持鲁棒精度的同时显著提升了训练速度。
摘要：
深度神经网络（DNNs）被用于解决许多领域中的广泛问题，包括自动驾驶汽车和医学成像等安全关键领域。然而，DNNs 容易受到对抗攻击的威胁。在过去的几年中，人们提出了许多方法，通过对抗训练来解决这个问题。几乎所有的方法都会为整个训练数据集生成对抗样本，从而大幅增加了训练时间。我们证明，通过仅使用一部分训练数据进行对抗训练，可以减少任何对抗训练算法的训练时间。为了选择这个子集，我们从训练数据中筛选出易受对抗攻击的样本。我们对所有训练样本执行简单的对抗攻击来筛选这个子集。在这种攻击中，我们向每个像素添加小的扰动，并向输入图像添加几条网格线。我们在易受对抗攻击的子集上执行对抗训练，并将其与在整个数据集上执行的普通训练相混合。我们的结果表明，当我们将这种与具体方法无关的方案插入 FGSM 时，在 MNIST 数据集上实现了 3.52 倍的加速，在 CIFAR-10 数据集上实现了 1.98 倍的加速，且鲁棒精度相当。我们还在最先进的 Free 对抗训练上测试了我们的方法，在 ImageNet 数据集上的训练时间实现了 1.2 倍的加速，且鲁棒精度仅有微小下降。
英文摘要：
Deep Neural Networks (DNNs) are being used to solve a wide range of problems in many domains including safety-critical domains like self-driving cars and medical imagery. DNNs suffer from vulnerability against adversarial attacks. In the past few years, numerous approaches have been proposed to tackle this problem by training networks using adversarial training. Almost all the approaches generate adversarial examples for the entire training dataset, thus increasing the training time drastically. We show that we can decrease the training time for any adversarial training algorithm by using only a subset of training data for adversarial training. To select the subset, we filter the adversarially-prone samples from the training data. We perform a simple adversarial attack on all training examples to filter this subset. In this attack, we add a small perturbation to each pixel and a few grid lines to the input image. We perform adversarial training on the adversarially-prone subset and mix it with vanilla training performed on the entire dataset. Our results show that when our method-agnostic approach is plugged into FGSM, we achieve a speedup of 3.52x on MNIST and 1.98x on the CIFAR-10 dataset with comparable robust accuracy. We also test our approach on state-of-the-art Free adversarial training and achieve a speedup of 1.2x in training time with a marginal drop in robust accuracy on the ImageNet dataset.
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。
