Synonym Substitution Attack

Synonym substitution attacks exploit the vulnerability of natural language processing (NLP) models to subtle word replacements, aiming to manipulate model outputs while preserving surface meaning. Current research focuses on improving the robustness of models against these attacks, exploring techniques like randomized smoothing, causal intervention, and synonym-aware pretraining, often within the context of black-box attack scenarios where model internals are hidden. Understanding and mitigating these attacks is crucial for ensuring the reliability and security of NLP systems across various applications, from machine translation to text classification.

Papers

August 29, 2023

A Classification-Guided Approach for Adversarial Attacks against Neural Machine Translation
Sahar Sadrizadeh, Ljiljana Dolamic, Pascal Frossard
Adversarial Attack Adversarial Example Neural Machine Translation Synonym Substitution Attack

July 31, 2023

Text-CRS: A Generalized Certified Robustness Framework against Textual Adversarial Attacks
Xinyu Zhang, Hanbin Hong, Yuan Hong, Peng Huang, Binghui Wang, Zhongjie Ba, Kui Ren
Textual Adversarial Attack Certified Robustness Synonym Substitution Attack Insertion Based Word Level Adversarial

July 24, 2023

Latent Code Augmentation Based on Stable Diffusion for Data-free Substitute Attacks
Mingwen Shao, Lingzhuang Meng, Yuanjian Qiao, Lixu Zhang, Wangmeng Zuo
GAN Model Stable Diffusion Latent Code Synonym Substitution Attack Code Augmentation

May 14, 2023

Watermarking Text Generated by Black-Box Language Models
Xi Yang, Kejiang Chen, Weiming Zhang, Chang Liu, Yuang Qi, Jie Zhang, Han Fang, Nenghai Yu
Watermarking Method Black Box Language Model Text Watermarking Watermarking Framework Synonym Substitution Attack

March 25, 2023

Sem4SAP: Synonymous Expression Mining From Open Knowledge Graph For Language Model Synonym-Aware Pretraining
Zhouhong Gu, Sihang Jiang, Wenhao Huang, Jiaqing Liang, Hongwei Feng, Yanghua Xiao
Language Model Knowledge Graph Pre Training Task Synonym Substitution Attack

October 6, 2022

Are Synonym Substitution Attacks Really Synonym Substitution Attacks?
Cheng-Han Chiang, Hung-yi Lee
Adversarial Sample Synonym Substitution Attack Word Substitution

May 24, 2022

Certified Robustness Against Natural Language Attacks by Causal Intervention
Haiteng Zhao, Chang Ma, Xinshuai Dong, Anh Tuan Luu, Zhi-Hong Deng, Hanwang Zhang
Native Robustness Adversarial Example Adversarial Vulnerability Causal Intervention Word Level Attack Synonym Substitution Attack

May 7, 2022

A Simple Yet Efficient Method for Adversarial Word-Substitute Attack
Tianle Li, Yi Yang
Adversarial Example Efficient Method Adversarial Question Synonym Substitution Attack

April 3, 2022

DST: Dynamic Substitute Training for Data-free Black-box Attack
Wenxuan Wang, Xuelin Qian, Yanwei Fu, Xiangyang Xue
Adversarial Example Substitute Model Synonym Substitution Attack Dynamic Training DST Det Data Free Black Box Attack

December 10, 2021

Batch Label Inference and Replacement Attacks in Black-Boxed Vertical Federated Learning
Yang Liu, Tianyuan Zou, Yan Kang, Wenhan Liu, Yuanqin He, Zhihao Yi, Qiang Yang
Homomorphic Encryption Vertical Federated Learning Clean Label Backdoor Attack Label Inference Attack Synonym Substitution Attack Batch Inference

Synonym Substitution Attack

Papers

A Classification-Guided Approach for Adversarial Attacks against Neural Machine Translation

Text-CRS: A Generalized Certified Robustness Framework against Textual Adversarial Attacks

Latent Code Augmentation Based on Stable Diffusion for Data-free Substitute Attacks

Watermarking Text Generated by Black-Box Language Models

Sem4SAP: Synonymous Expression Mining From Open Knowledge Graph For Language Model Synonym-Aware Pretraining

Are Synonym Substitution Attacks Really Synonym Substitution Attacks?

Certified Robustness Against Natural Language Attacks by Causal Intervention

A Simple Yet Efficient Method for Adversarial Word-Substitute Attack

DST: Dynamic Substitute Training for Data-free Black-box Attack

Batch Label Inference and Replacement Attacks in Black-Boxed Vertical Federated Learning