arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.07327cs.LG
收藏

用于对抗训练的引导插值

Guided Interpolation for Adversarial Training

Zhejiang University(浙江大学);RIKEN Center for Advanced Intelligence Project (AIP)(理化学研究所先进智能项目中心);Shandong University(山东大学);The University of Tokyo(东京大学)

Chen Chen, Jingfeng Zhang, Xilie Xu, Tianlei Hu, Gang Niu, Gang Chen, Masashi Sugiyama

摘要:

为了增强对抗性鲁棒性,对抗性训练在其自然数据生成的对抗性变体上学习深度神经网络。然而,随着训练的进行,训练数据变得越来越不可攻击,破坏了鲁棒性增强。一个简单的补救办法是合并更多的训练数据,但有时会产生无法承受的成本。在本文中,为了缓解这个问题,我们提出了引导插值框架(GIF):在每个时期,GIF采用前一个时期的Meta信息来指导数据的插值。与vanilla mixup相比,GIF能够提供更高比例的可攻击数据,这有利于增强鲁棒性;同时,GIF也缓解了模型类间的线性行为,线性行为有利于泛化,但不利于鲁棒性。因此,GIF鼓励模型在每个类的聚类中进行不变的预测。实验表明,GIF确实可以增强各种对抗训练方法和各种数据集的对抗鲁棒性。

英文摘要:

To enhance adversarial robustness, adversarial training learns deep neural networks on the adversarial variants generated by their natural data. However, as the training progresses, the training data becomes less and less attackable, undermining the robustness enhancement. A straightforward remedy is to incorporate more training data, but sometimes incurring an unaffordable cost. In this paper, to mitigate this issue, we propose the guided interpolation framework (GIF): in each epoch, the GIF employs the previous epoch's meta information to guide the data's interpolation. Compared with the vanilla mixup, the GIF can provide a higher ratio of attackable data, which is beneficial to the robustness enhancement; it meanwhile mitigates the model's linear behavior between classes, where the linear behavior is favorable to generalization but not to the robustness. As a result, the GIF encourages the model to predict invariantly in the cluster of each class. Experiments demonstrate that the GIF can indeed enhance adversarial robustness on various adversarial training methods and various datasets.

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。

↑