Adversarial Purification

Adversarial purification aims to remove malicious perturbations from data inputs, thereby improving the robustness of machine learning models against adversarial attacks without modifying the underlying classifier. Current research heavily utilizes diffusion models and generative adversarial networks (GANs), often incorporating techniques like classifier guidance and adaptive noise control to enhance purification accuracy and efficiency, particularly for image and text data. This field is crucial for bolstering the reliability and security of AI systems across diverse applications, from mobile device security to intrusion detection and power system monitoring, by mitigating the impact of increasingly sophisticated adversarial attacks.

Papers

October 28, 2023

Purify++: Improving Diffusion-Purification with Advanced Diffusion Models and Control of Randomness
Boya Zhang, Weijian Luo, Zhihua Zhang
Adversarial Attack External Control Inherent Randomness Adversarial Purification Diffusion Purification

September 19, 2023

Language Guided Adversarial Purification
Himanshu Singh, A V Subramanyam
Adversarial Attack Adversarial Purification Adversarial Defense Adversarial Text Purification

August 29, 2023

Advancing Adversarial Robustness Through Adversarial Logit Update
Hao Xuan, Peican Zhu, Xingyu Li
Adversarial Attack Adversarial Training Adversarial Robustness Adversarial Sample Adversarial Purification Successful Adversarial Attack

August 28, 2023

DiffSmooth: Certifiably Robust Learning via Diffusion Models and Local Smoothing
Jiawei Zhang, Zhongzhu Chen, Huan Zhang, Chaowei Xiao, Bo Li
Diffusion Model Native Robustness Robust Learning Adversarial Purification Certified Robustness Differentiable ML Pipeline

July 31, 2023

Universal Adversarial Defense in Remote Sensing Based on Pre-trained Denoising Diffusion Models
Weikang Yu, Yonghao Xu, Pedram Ghamisi
Remote Sensing Denoising Diffusion Model Adversarial Purification Universal Adversarial

May 27, 2023

Rapid Plug-in Defenders
Kai Wu, Yujian Betterest Li, Jian Lou, Xiaoyu Zhang, Handing Wang, Jing Liu
Adversarial Attack Adversarial Example Transformer Based Adversarial Training Adversarial Purification

May 25, 2023

Carefully Blending Adversarial Training and Purification Improves Adversarial Robustness
Emanuele Ballarin, Alessio Ansuini, Luca Bortolussi
Adversarial Training Adversarial Robustness Adversarial Learning Generative Adversarial Adversarial DEfense Adversarial Purification

March 2, 2023

Defending against Adversarial Audio via Diffusion Model
Shutong Wu, Jiongxiao Wang, Wei Ping, Weili Nie, Chaowei Xiao
Diffusion Model Adversarial Attack Adversarial Training Adversarial Purification Adversarial Audio Audio Adversarial Example

February 3, 2023

Leveraging Contaminated Datasets to Learn Clean-Data Distribution with Purified Generative Adversarial Networks
Bowen Tian, Qinliang Su, Jianxing Yu
Generative Adversarial Network Training Data GAN Model Adversarial Learning Generative Question Adversarial Purification

October 26, 2022

Adversarial Purification with the Manifold Hypothesis
Zhaoyuan Yang, Zhiwei Xu, Jing Zhang, Richard Hartley, Peter Tu
Adversarial Example Adversarial Robustness Variational Autoencoders Manifold Learning Adversarial Purification Manifold Hypothesis

May 16, 2022

Diffusion Models for Adversarial Purification
Weili Nie, Brandon Guo, Yujia Huang, Chaowei Xiao, Arash Vahdat, Anima Anandkumar
Diffusion Model Adversarial Example Adversarial Training Adversarial Perturbation Adversarial Purification

March 27, 2022

Text Adversarial Purification as Defense against Adversarial Attacks
Linyang Li, Demin Song, Xipeng Qiu
Adversarial Attack Adversarial Purification Textual Adversarial Attack Adversarial Text Purification