Adversarial attacks induce misclassification by introducing subtle perturbations. Recently, diffusion models are applied to the image classifiers to improve adversarial robustness through adversarial training or by purifying adversarial noise. However, diffusion-based adversarial training often encounters convergence challenges and high computational expenses. Additionally, diffusion-based purification inevitably causes data shift and is deemed susceptible to stronger adaptive attacks. To tackle these issues, we propose the Truth Maximization Diffusion Classifier (TMDC), a generative Bayesian classifier that builds upon pre-trained diffusion models and the Bayesian theorem. Unlike data-driven classifiers, TMDC, guided by Bayesian principles, utilizes the conditional likelihood from diffusion models to determine the class probabilities of input images, thereby insulating against the influences of data shift and the limitations of adversarial training. Moreover, to enhance TMDC's resilience against more potent adversarial attacks, we propose an optimization strategy for diffusion classifiers. This strategy involves post-training the diffusion model on perturbed datasets with ground-truth labels as conditions, guiding the diffusion model to learn the data distribution and maximizing the likelihood under the ground-truth labels. The proposed method achieves state-of-the-art performance on the CIFAR10 dataset against heavy white-box attacks and strong adaptive attacks. Specifically, TMDC achieves robust accuracies of 82.81% against $l_{\infty}$ norm-bounded perturbations and 86.05% against $l_{2}$ norm-bounded perturbations, respectively, with $ε=0.05$.
The Linear Geometry of Interpretable Tokens: Jailbreaking Attacks and Defenses for Unlearned Diffusion Models
Department of Electrical Engineering & Computer Science, University of Michigan(电气工程与计算机科学系,密歇根大学);Department of Computer Science, Michigan State University(计算机科学系,密歇根州立大学)
AlignVTOFF: Texture-Spatial Feature Alignment for High-Fidelity Virtual Try-Off
Sino-European School of Technology, Shanghai University(上海大学 sino-欧洲技术学院);National Demonstration Center for Experimental Engineering Training Education, Shanghai University(上海大学国家级实验工程培训教育示范中心)
ModaFlow: Modality-Aware Flow Matching for High-Fidelity Virtual Try-On
South China University of Technology(华南理工大学);Universitat de Barcelona(巴塞罗那大学);Computer Vision Center(计算机视觉中心);Guangdong Provincial Key Laboratory of Multimodal Big Data Intelligent Analysis(广东省多模态大数据智能分析重点实验室)
An Adaptive Black-box Defense against Trojan Attacks (TrojDef)
New Jersey Institute of Technology(新泽西理工学院);Jordan University of Science & Technology(约旦科技大学);Qatar Computing Research Institute(卡塔尔计算研究所);HBKU(哈马德·本·哈利法大学)
REFINE: Inversion-Free Backdoor Defense via Model Reprogramming
Zhejiang University(浙江大学);Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security(杭州高新区(滨江)区块链与数据安全研究院);Nanyang Technological University(南洋理工大学);IBM Research(IBM研究院)
PAD: Patch-Agnostic Defense against Adversarial Patch Attacks
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院);School of Cyber Science and Technology, Shenzhen Campus of Sun Yat-sen University(中山大学深圳校区网络空间安全学院)
AD-SRF: Attack-Defense Semantic Reinforcement for Multi-Modal Image Fusion
Electronic Information School, Wuhan University(武汉大学电子信息学院);School of Robotics, Wuhan University(武汉大学机器人学院);Suzhou Institute, Wuhan University(武汉大学苏州研究院)
Hyper Adversarial Tuning for Boosting Adversarial Robustness of Pretrained Large Vision Models
Zhejiang University(浙江大学);Shanghai Maritime University(上海海事大学);Zhejiang Gongshang University(浙江工商大学);Nanjing University of Information Science and Technology(南京信息工程大学)
School of Computer Science, Northwestern Polytechnical University,National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology(计算机科学学院,西北工业大学,集成空天地海大数据应用技术国家工程实验室);School of Automation, Northwestern Polytechnical University,the BRain and Artificial INtelligence Lab(自动化学院,西北工业大学,脑与人工智能实验室)
Undress to Redress: A Training-Free Framework for Virtual Try-On
Zhiying Li 1(李子莹1);Junhao Wu 3(吴俊浩3);Yeying Jin 4(金悦莹4);Daiheng Gao 5(高代恒5);Yun Ji 2(纪云2);Kaichuan Kong 1(孔开传1);Lei Yu 6(于磊6);Hao Xu 7(徐浩7);Kai Chen 7(陈凯7);Bruce Gu 8(古布鲁斯8);Nana Wang 9(王娜9);Zhaoxin Fan 2(范朝新2)