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 2311.06647cs.CL
收藏

鲁棒文本分类:分析基于原型的网络

Robust Text Classification: Analyzing Prototype-Based Networks

Information Sciences Institute, University of Southern California(南加州大学信息科学研究所);Department of Computer Science, University of Southern California(南加州大学计算机科学系);Vrije Universiteit Amsterdam(阿姆斯特丹自由大学);NEC Laboratories Europe(NEC欧洲实验室);CAIR, Ss. Cyril and Methodius University(圣西里尔和美多德大学CAIR)

Zhivar Sourati, Darshan Deshpande, Filip Ilievski, Kiril Gashteovski, Sascha Saralajew

AI总结:

本文研究基于原型的网络在文本分类中的对抗鲁棒性,证明其作为语言模型架构变体在定向与静态攻击下更鲁棒,并分析可解释性及聚类紧密度的影响。

摘要:

下游应用通常要求文本分类模型既准确又鲁棒。尽管最先进的语言模型(LMs)的准确率已接近人类水平,但它们在现实世界中存在的噪声数据上往往会出现性能下降。这种鲁棒性缺失可能令人担忧,因为即使文本中与目标任务无关的微小扰动,也会导致分类器错误地改变其预测。一种潜在解决方案是基于原型的网络(PBNs)家族,该网络根据样本与某一类别的原型样本(prototypes)的相似性对样本进行分类,并且已在计算机视觉任务中被证明对噪声具有鲁棒性。本文研究在定向对抗攻击和静态对抗攻击两种设置下,PBNs的鲁棒性是否能够迁移到文本分类任务。结果表明,PBNs作为普通LMs的一种单纯架构变体,在定向和静态两种设置下都比普通LMs具有更强的鲁棒性。我们展示了PBNs的可解释性如何帮助理解其鲁棒性。最后,消融研究揭示了PBNs的鲁棒性对训练阶段聚类严格程度的敏感性:更紧密的聚类会导致PBNs鲁棒性更差。

英文摘要:

Downstream applications often require text classification models to be accurate and robust. While the accuracy of the state-of-the-art Language Models (LMs) approximates human performance, they often exhibit a drop in performance on noisy data found in the real world. This lack of robustness can be concerning, as even small perturbations in the text, irrelevant to the target task, can cause classifiers to incorrectly change their predictions. A potential solution can be the family of Prototype-Based Networks (PBNs) that classifies examples based on their similarity to prototypical examples of a class (prototypes) and has been shown to be robust to noise for computer vision tasks. In this paper, we study whether the robustness properties of PBNs transfer to text classification tasks under both targeted and static adversarial attack settings. Our results show that PBNs, as a mere architectural variation of vanilla LMs, offer more robustness compared to vanilla LMs under both targeted and static settings. We showcase how PBNs' interpretability can help us to understand PBNs' robustness properties. Finally, our ablation studies reveal the sensitivity of PBNs' robustness to how strictly clustering is done in the training phase, as tighter clustering results in less robust PBNs.

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。

↑