AdvCodeMix: Adversarial Attack on Code-Mixed Data

Abstract

Research on adversarial attacks are becoming widely popular in the recent years. One of the unexplored areas where prior research is lacking is the effect of adversarial attacks on code-mixed data. Therefore, in the present work, we have explained the first generalized framework on text perturbation to attack code-mixed classification models in a black-box setting. We rely on various perturbation techniques that preserve the semantic structures of the sentences and also obscure the attacks from the perception of a human user. The present methodology leverages the importance of a token to decide where to attack by employing various perturbation strategies. We test our strategies on various sentiment classification models trained on Bengali-English and Hindi-English code-mixed datasets, and reduce their F1-scores by nearly 51 % and 53 % respectively, which can be further reduced if a larger number of tokens are perturbed in a given sentence.

arXiv commentAccepted to CODS-COMAD 2022

References in corpus (10)

• Character-level Convolutional Networks for Text Classification
• Towards Deep Learning Models Resistant to Adversarial Attacks
• MagNet: a Two-Pronged Defense against Adversarial Examples
• How multilingual is Multilingual BERT?
• Sentiment Analysis of Code-Mixed Indian Languages: An Overview of SAIL_Code-Mixed Shared Task @ICON-2017
• BERT-ATTACK: Adversarial Attack Against BERT Using BERT
• Preparing Bengali-English Code-Mixed Corpus for Sentiment Analysis of Indian Languages
• Natural Backdoor Attack on Text Data
• kk2018 at SemEval-2020 Task 9: Adversarial Training for Code-Mixing Sentiment Classification
• Code-Mixing on Sesame Street: Dawn of the Adversarial Polyglots