Adversarial Text

Adversarial text research focuses on creating and defending against text inputs designed to deceive natural language processing (NLP) models, often by subtly altering wording while maintaining semantic similarity to a human reader. Current research emphasizes developing more effective attack methods, particularly those leveraging multi-agent systems, reinforcement learning, and diffusion models, as well as improving defenses through techniques like adversarial training and noise augmentation. This field is crucial for enhancing the robustness and trustworthiness of NLP systems across diverse applications, from automated essay scoring to autonomous vehicle navigation and large language model safety.

Papers

October 7, 2022

LLMEffiChecker: Understanding and Testing Efficiency Degradation of Large Language Models
Xiaoning Feng, Xiaohong Han, Simin Chen, Wei Yang
Human Understanding Black Box Adversarial Text Computational Efficiency Robust Algorithm Test Time Optimization

September 15, 2022

Explicit Tradeoffs between Adversarial and Natural Distributional Robustness
Mazda Moayeri, Kiarash Banihashem, Soheil Feizi
Adversarial Training Adversarial Robustness Adversarial Text Adversarial Vulnerability Distributional Robustness

September 10, 2022

Adversarial Learning-based Stance Classifier for COVID-19-related Health Policies
Feng Xie, Zhong Zhang, Xuechen Zhao, Haiyang Wang, Jiaying Zou, Lei Tian, Bin Zhou, Yusong Tan
Adversarial Learning Stance Detection Adversarial Text

August 31, 2022

Be Your Own Neighborhood: Detecting Adversarial Example by the Neighborhood Relations Built on Self-Supervised Learning
Zhiyuan He, Yijun Yang, Pin-Yu Chen, Qiang Xu, Tsung-Yi Ho
Deep Neural Network Self Supervised Learning Adversarial Example Adversarial Text Thy Neighbor Adaptive Attack Adversarial Example Detection

June 16, 2022

May 3, 2022

SemAttack: Natural Textual Attacks via Different Semantic Spaces
Boxin Wang, Chejian Xu, Xiangyu Liu, Yu Cheng, Bo Li
Adversarial Text Semantic Space Textual Adversarial Attack Textual Attack

April 12, 2022

Examining the Proximity of Adversarial Examples to Class Manifolds in Deep Networks
Štefan Pócoš, Iveta Bečková, Igor Farkaš
Deep Neural Network Adversarial Example Deep Network Internal Representation Adversarial Text Class Manifold Hidden Layer Proximity Search

March 25, 2022

Analysis of OODA Loop based on Adversarial for Complex Game Environments
Xiangri Lu, Hongbin Ma, Zhanqing Wang
General Analysis Adversarial Text Rapid Countermeasure Game Environment Hierarchical Game

March 24, 2022

Addressing Missing Sources with Adversarial Support-Matching
Thomas Kehrenberg, Myles Bartlett, Viktoriia Sharmanska, Novi Quadrianto
Machine Learning Model Adversarial Text Semi Supervised Clustering Unlabeled Dataset Representation Bias

March 19, 2022

Perturbations in the Wild: Leveraging Human-Written Text Perturbations for Realistic Adversarial Attack and Defense
Thai Le, Jooyoung Lee, Kevin Yen, Yifan Hu, Dongwon Lee
Wild Challenge Adversarial Text Additive Perturbation Toxic Text Realistic Adversarial

March 17, 2022

ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection
Thomas Hartvigsen, Saadia Gabriel, Hamid Palangi, Maarten Sap, Dipankar Ray, Ece Kamar
Hate Speech Detection Adversarial Text Toxicity Detection Toxic Language Toxic Text Toxic Speech Detection

February 26, 2022

February 4, 2022

Pixle: a fast and effective black-box attack based on rearranging pixels
Jary Pomponi, Simone Scardapane, Aurelio Uncini
Adversarial Attack Black Box Adversarial Sample Tetromino Pixel Adversarial Text

January 26, 2022

Generative Trees: Adversarial and Copycat
Richard Nock, Mathieu Guillame-Bert
Generative Adversarial Network Feature Imitation Adversarial Text GAN Loss

January 21, 2022

Identifying Adversarial Attacks on Text Classifiers
Zhouhang Xie, Jonathan Brophy, Adam Noack, Wencong You, Kalyani Asthana, Carter Perkins, Sabrina Reis, Sameer Singh, Daniel Lowd
Adversarial Attack Adversarial Text Text Classifier Attack Detection

December 29, 2021

Repairing Adversarial Texts through Perturbation
Guoliang Dong, Jingyi Wang, Jun Sun, Sudipta Chattopadhyay, Xinyu Wang, Ting Dai, Jie Shi, Jin Song Dong
Adversarial Perturbation Adversarial Text Small Perturbation

November 16, 2021

Adversarially Constructed Evaluation Sets Are More Challenging, but May Not Be Fair
Jason Phang, Angelica Chen, William Huang, Samuel R. Bowman
Adversarial Text Evaluation Datasets Evaluation Set Challenging Dataset Adversarial Datasets Adversary Model

Adversarial Text

Papers

LLMEffiChecker: Understanding and Testing Efficiency Degradation of Large Language Models

Explicit Tradeoffs between Adversarial and Natural Distributional Robustness

Adversarial Learning-based Stance Classifier for COVID-19-related Health Policies

Be Your Own Neighborhood: Detecting Adversarial Example by the Neighborhood Relations Built on Self-Supervised Learning

Gradient-Based Adversarial and Out-of-Distribution Detection

A Closer Look at Smoothness in Domain Adversarial Training

SemAttack: Natural Textual Attacks via Different Semantic Spaces

Examining the Proximity of Adversarial Examples to Class Manifolds in Deep Networks

Analysis of OODA Loop based on Adversarial for Complex Game Environments

Addressing Missing Sources with Adversarial Support-Matching

Perturbations in the Wild: Leveraging Human-Written Text Perturbations for Realistic Adversarial Attack and Defense

ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection

BioADAPT-MRC: Adversarial Learning-based Domain Adaptation Improves Biomedical Machine Reading Comprehension Task

DGSS : Domain Generalized Semantic Segmentation using Iterative Style Mining and Latent Representation Alignment

Pixle: a fast and effective black-box attack based on rearranging pixels

Generative Trees: Adversarial and Copycat

Identifying Adversarial Attacks on Text Classifiers

Repairing Adversarial Texts through Perturbation

Adversarially Constructed Evaluation Sets Are More Challenging, but May Not Be Fair