Unlearnable examples (ULEs) aim to protect data from unauthorized usage for training DNNs. Error-minimizing noise, which is injected to clean data, is one of the most successful methods for preventing DNNs from giving correct predictions on incoming new data. Nonetheless, under specific training strategies such as adversarial training, the unlearnability of error-minimizing noise will severely degrade. In addition, the transferability of error-minimizing noise is inherently limited by the mismatch between the generator model and the targeted learner model. In this paper, we investigate the mechanism of unlearnable examples and propose a novel model-free method, named \emph{One-Pixel Shortcut}, which only perturbs a single pixel of each image and makes the dataset unlearnable. Our method needs much less computational cost and obtains stronger transferability and thus can protect data from a wide range of different models. Based on this, we further introduce the first unlearnable dataset called CIFAR-10-S, which is indistinguishable from normal CIFAR-10 by human observers and can serve as a benchmark for different models or training strategies to evaluate their abilities to extract critical features from the disturbance of non-semantic representations. The original error-minimizing ULEs will lose efficiency under adversarial training, where the model can get over 83\% clean test accuracy. Meanwhile, even if adversarial training and strong data augmentation like RandAugment are applied together, the model trained on CIFAR-10-S cannot get over 50\% clean test accuracy.
翻译:不可忽略的例子(ULEs)旨在保护数据不被未经授权地用于培训 DNNs 。 错误最小化噪音被注入清洁数据, 是防止 DNNs对收到的新数据作出正确预测的最成功方法之一 。 尽管如此, 在对抗性培训等具体培训战略下, 错误最小化噪音的不可忽略性会严重降低 。 此外, 最小化噪音的可转移性受到发电机模型与目标学习者模型之间的不匹配的内在限制 。 在本文中, 我们调查不可忽略的准确性实例机制, 并提出一种新型的无精确性方法, 名为 emph{ One- Pixel 快捷}, 它只能渗透每个图像的单一像素位, 使数据集无法读取 。 我们的方法需要更低得多的计算成本, 获得更强的可转移性, 从而保护数据不受多种不同模型的影响 。 基于此, 我们进一步介绍第一个不可忽略的数据集, 称为 CFAR- X- S- S, 它叫做 CHR- 10- S,, 它是一个新型的不易读性模型, 它比正常的测试模型, 可以比正常的 RR- treal- trestal- trem- train train view destal view view view view view view view view view view viewal viewal viewal view viol violmmmmmmmmol viol vi vi vi viewmational viewmational view viewmational viewdal views viewdal viewdal views 。