Clean Label Attack

Clean-label backdoor attacks are a stealthy form of data poisoning that manipulates machine learning models without altering training labels, making them difficult to detect. Current research focuses on developing more effective attack strategies, particularly those requiring minimal knowledge of the training data, and exploring defenses against these attacks across various model architectures, including neural networks, gradient boosting models, and language models like BERT. The ability to subtly inject malicious behavior into models highlights significant vulnerabilities in machine learning systems, impacting the trustworthiness of AI in security-critical applications and necessitating the development of robust detection and mitigation techniques.

Papers

September 13, 2024

Clean Label Attacks against SLU Systems
Henry Li Xinyuan, Sonal Joshi, Thomas Thebaud, Jesus Villalba, Najim Dehak, Sanjeev Khudanpur
Data Poisoning Attack Gradient Based Attack Backdoor Poisoning Attack End 2 End Clean Label Backdoor Attack Clean Label Attack

July 15, 2024

Wicked Oddities: Selectively Poisoning for Effective Clean-Label Backdoor Attacks
Quang H. Nguyen, Nguyen Ngoc-Hieu, The-Anh Ta, Thanh Nguyen-Tang, Kok-Seng Wong, Hoang Thanh-Tung, Khoa D. Doan
Adversarial Attack Backdoor Attack Attack Success Rate Clean Label Backdoor Attack Clean Label Attack

July 11, 2024

Model-agnostic clean-label backdoor mitigation in cybersecurity environments
Giorgio Severi, Simona Boboila, John Holodnak, Kendra Kratkiewicz, Rauf Izmailov, Michael J. De Lucia, Alina Oprea
Backdoor Defense Threat Model Training Time Attack Clean Label Backdoor Attack Clean Label Attack Label Poisoning Attack Cybersecurity Environment

May 9, 2024

Hard Work Does Not Always Pay Off: Poisoning Attacks on Neural Architecture Search
Zachary Coalson, Huazheng Wang, Qingyun Wu, Sanghyun Hong
Neural Architecture Search Poisoning Attack Manual Effort Clean Label Attack Adversarial Discriminator

March 30, 2024

Shortcuts Arising from Contrast: Effective and Covert Clean-Label Attacks in Prompt-Based Learning
Xiaopeng Xie, Ming Yan, Xiwen Zhou, Chenlong Zhao, Suli Wang, Yong Zhang, Joey Tianyi Zhou
Backdoor Attack Prompt Based Learning Shortcut Feature Clean Label Attack Label Poisoning Attack

March 18, 2024

Diffusion Denoising as a Certified Defense against Clean-label Poisoning
Sanghyun Hong, Nicholas Carlini, Alexey Kurakin
Adversarial Robustness Adversarial Perturbation Denoising Diffusion Clean Label Attack Certified Defense Label Poisoning Attack Label Poisoning

October 28, 2023

Large Language Models Are Better Adversaries: Exploring Generative Clean-Label Backdoor Attacks Against Text Classifiers
Wencong You, Zayd Hammoudeh, Daniel Lowd
Adversarial Example Backdoor Attack Attack Success Rate Adversary Agent Text Classifier Textual Backdoor Attack Clean Label Backdoor Attack Clean Label Attack

October 23, 2023

Attention-Enhancing Backdoor Attacks Against BERT-based Models
Weimin Lyu, Songzhu Zheng, Lu Pang, Haibin Ling, Chao Chen
Backdoor Attack BERT Based Trojan Detection Textual Backdoor Attack Clean Label Attack

August 18, 2023

DFB: A Data-Free, Low-Budget, and High-Efficacy Clean-Label Backdoor Attack
Binhao Ma, Jiahui Wang, Dejun Wang, Bo Meng
Clean Label Backdoor Attack Clean Label Attack Low Budget Semiconductor Laser

May 31, 2023

Adversarial Clean Label Backdoor Attacks and Defenses on Text Classification Systems
Ashim Gupta, Amrith Krishna
Text Classification Clean Label Backdoor Attack Text Attack Clean Label Attack Clean Label

February 22, 2023

ASSET: Robust Backdoor Data Detection Across a Multiplicity of Deep Learning Paradigms
Minzhou Pan, Yi Zeng, Lingjuan Lyu, Xue Lin, Ruoxi Jia
Backdoor Attack Backdoor Defense Model Multiplicity Clean Label Attack

January 5, 2023

Silent Killer: A Stealthy, Clean-Label, Black-Box Backdoor Attack
Tzvi Lederer, Gallil Maimon, Lior Rokach
Backdoor Poisoning Attack Clean Label Attack Label Poisoning Poison Detection

April 11, 2022

Narcissus: A Practical Clean-Label Backdoor Attack with Limited Information
Yi Zeng, Minzhou Pan, Hoang Anh Just, Lingjuan Lyu, Meikang Qiu, Ruoxi Jia
Backdoor Attack Backdoor Trigger Clean Label Backdoor Attack Limited Information Clean Label Attack Malicious Data