arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1704.08006cs.CRcs.LG
收藏

深度文本分类可以被愚弄

Deep Text Classification Can be Fooled

Renmin University of China(中国人民大学);School of Information(信息学院);Key laboratory of Data Engineering and Knowledge Engineering, MOE(教育部数据工程与知识工程重点实验室)

Bin Liang, Hongcheng Li, Miaoqiang Su, Pan Bian, Xirong Li, Wenchang Shi

摘要:

在本文中,我们提出了一种有效的方法来制作文本对抗样本,揭示了一个重要但被低估的事实,即基于DNN的文本分类器也容易受到对抗样本攻击。具体来说,面对不同的对抗场景,通过计算输入的成本梯度(白盒攻击)或生成一系列被遮挡的测试样本(黑盒攻击)来识别对分类重要的文本项。基于这些项,我们设计了三种扰动策略,即插入,修改和删除,以生成对抗样本。实验结果表明,我们的方法生成的对抗性样本可以成功地欺骗最先进的字符级和单词级基于DNN的文本分类器。对抗样本可以被扰动到任何期望的类别,而不会损害它们的效用。同时,引入的扰动难以被感知。

英文摘要:

In this paper, we present an effective method to craft text adversarial samples, revealing one important yet underestimated fact that DNN-based text classifiers are also prone to adversarial sample attack. Specifically, confronted with different adversarial scenarios, the text items that are important for classification are identified by computing the cost gradients of the input (white-box attack) or generating a series of occluded test samples (black-box attack). Based on these items, we design three perturbation strategies, namely insertion, modification, and removal, to generate adversarial samples. The experiment results show that the adversarial samples generated by our method can successfully fool both state-of-the-art character-level and word-level DNN-based text classifiers. The adversarial samples can be perturbed to any desirable classes without compromising their utilities. At the same time, the introduced perturbation is difficult to be perceived.

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。

↑