arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2101.12372cs.LG
收藏

成本敏感类的对抗学习

Adversarial Learning with Cost-Sensitive Classes

Shenzhen University(深圳大学)

Haojing Shen, Sihong Chen, Ran Wang, Xizhao Wang

摘要:

在对抗性学习中,有必要提高某些特殊类的性能,或者特别保护它们免受攻击。本文提出了一种结合成本敏感分类和对抗学习的框架,训练一个模型,可以区分受保护和不受保护的类,使受保护的类不容易受到对抗的例子。在这个框架中,我们发现了深度神经网络训练过程中的一个有趣现象,称为Min-Max特性,即卷积层中大多数参数的绝对值接近于零,而少数参数的绝对值明显变大。在此基础上,从随机分布的角度对Min-Max性质进行了分析,并进一步建立了一种新的对抗性样本防御模型,以提高对抗性样本的鲁棒性。所建模型的一个优点是它的性能优于标准模型,并且可以与对抗训练相结合,以实现更好的性能。实验证实,关于所有类的平均准确度,我们的模型几乎是相同的现有模型时,攻击没有发生,是优于现有的模型时,攻击发生。具体而言,关于受保护类的准确性,当攻击发生时,所提出的模型比现有的模型好得多。

英文摘要:

It is necessary to improve the performance of some special classes or to particularly protect them from attacks in adversarial learning. This paper proposes a framework combining cost-sensitive classification and adversarial learning together to train a model that can distinguish between protected and unprotected classes, such that the protected classes are less vulnerable to adversarial examples. We find in this framework an interesting phenomenon during the training of deep neural networks, called Min-Max property, that is, the absolute values of most parameters in the convolutional layer approach zero while the absolute values of a few parameters are significantly larger becoming bigger. Based on this Min-Max property which is formulated and analyzed in a view of random distribution, we further build a new defense model against adversarial examples for adversarial robustness improvement. An advantage of the built model is that it performs better than the standard one and can combine with adversarial training to achieve an improved performance. It is experimentally confirmed that, regarding the average accuracy of all classes, our model is almost as same as the existing models when an attack does not occur and is better than the existing models when an attack occurs. Specifically, regarding the accuracy of protected classes, the proposed model is much better than the existing models when an attack occurs.

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。

↑