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 2209.07735cs.CV
收藏

通过离散对抗训练增强视觉表示

Enhance the Visual Representation via Discrete Adversarial Training

Alibaba Group(阿里巴巴集团);Zhejiang University(浙江大学);EPFL(洛桑联邦理工学院)

Xiaofeng Mao, Yuefeng Chen, Ranjie Duan, Yao Zhu, Gege Qi, Shaokai Ye, Xiaodan Li, Rong Zhang, Hui Xue

摘要:

对抗性训练(AT)被公认为对抗对抗性样本的最有效方法之一,但它在很大程度上会损害标准性能,因此在工业规模生产和应用中的实用性有限。令人惊讶的是,这种现象在自然语言处理(NLP)任务中完全相反,AT甚至可以受益于泛化。我们注意到AT在NLP任务中的优点可以来自离散和符号输入空间。为了借鉴NLP式AT的优点,我们提出了离散对抗训练(DAT)。DAT利用VQGAN将图像数据改造为离散的文本类输入,即视觉单词。然后,它最大限度地减少这种离散图像的符号对抗扰动的最大风险。作为一种即插即用的增强视觉表现的技术,DAT在图像分类、目标检测和自监督学习等多个任务上都取得了显著的改进。特别是,使用Masked Auto-Encoding(MAE)进行预训练并通过DAT进行微调而无需额外数据的模型可以在ImageNet-C上获得31.40 mCE,在Stylized-ImageNet上获得32.77%的top-1准确率,构建了最先进的模型。代码将在https://github.com/alibaba/easyrobust上提供。

英文摘要:

Adversarial Training (AT), which is commonly accepted as one of the most effective approaches defending against adversarial examples, can largely harm the standard performance, thus has limited usefulness on industrial-scale production and applications. Surprisingly, this phenomenon is totally opposite in Natural Language Processing (NLP) task, where AT can even benefit for generalization. We notice the merit of AT in NLP tasks could derive from the discrete and symbolic input space. For borrowing the advantage from NLP-style AT, we propose Discrete Adversarial Training (DAT). DAT leverages VQGAN to reform the image data to discrete text-like inputs, i.e. visual words. Then it minimizes the maximal risk on such discrete images with symbolic adversarial perturbations. We further give an explanation from the perspective of distribution to demonstrate the effectiveness of DAT. As a plug-and-play technique for enhancing the visual representation, DAT achieves significant improvement on multiple tasks including image classification, object detection and self-supervised learning. Especially, the model pre-trained with Masked Auto-Encoding (MAE) and fine-tuned by our DAT without extra data can get 31.40 mCE on ImageNet-C and 32.77% top-1 accuracy on Stylized-ImageNet, building the new state-of-the-art. The code will be available at https://github.com/alibaba/easyrobust.

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。

↑