SemanticAdv: Generating Adversarial Examples via Attribute-conditional Image Editing

2019/06/19 by Haonan Qiu, Chaowei Xiao, Qiu, Haonan +9 · 1 citation
Computer Science · Medicine · #Advanced Neural Network Applications #Adversarial Robustness in Machine Learning #COVID-19 diagnosis using AI #Computer Vision and Pattern Recognition (cs.CV) #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Machine Learning (cs.LG) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.1906.07927

openalex publication_date 2019/06/19 · openalex created_date 2019/06/27 · openalex updated_date 2026/07/28

Abstract

Deep neural networks (DNNs) have achieved great success in various applications due to their strong expressive power. However, recent studies have shown that DNNs are vulnerable to adversarial examples which are manipulated instances targeting to mislead DNNs to make incorrect predictions. Currently, most such adversarial examples try to guarantee "subtle perturbation" by limiting the L_p norm of the perturbation. In this paper, we aim to explore the impact of semantic manipulation on DNNs predictions by manipulating the semantic attributes of images and generate "unrestricted adversarial examples". In particular, we propose an algorithm SemanticAdv which leverages disentangled semantic factors to generate adversarial perturbation by altering controlled semantic attributes to fool the learner towards various "adversarial" targets. We conduct extensive experiments to show that the semantic based adversarial examples can not only fool different learning tasks such as face verification and landmark detection, but also achieve high targeted attack success rate against real-world black-box services such as Azure face verification service based on transferability. To further demonstrate the applicability of SemanticAdv beyond face recognition domain, we also generate semantic perturbations on street-view images. Such adversarial examples with controlled semantic manipulation can shed light on further understanding about vulnerabilities of DNNs as well as potential defensive approaches.

Citations

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Auto-Encoding Variational Bayes
• ImageNet classification with deep convolutional neural networks
• Pixel Recurrent Neural Networks
• Generative Adversarial Text to Image Synthesis
• DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
• A study of the effect of JPG compression on adversarial images
• Conditional Image Synthesis With Auxiliary Classifier GANs
• Learning Hierarchical Semantic Image Manipulation through Structured Representations
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Wasserstein Adversarial Examples via Projected Sinkhorn Iterations
• Generative Visual Manipulation on the Natural Image Manifold
• Generative Face Completion
• Conditional Image Generation with PixelCNN Decoders
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Cited by

• SCOOTER: A Human Evaluation Framework for Unrestricted Adversarial Examples

Related

• Semantic Adversarial Attacks: Parametric Transformations That Fool Deep\n Classifiers
• Transcending Adversarial Perturbations: Manifold-Aided Adversarial Examples with Legitimate Semantics
• Semantic Adversarial Examples
• Unrestricted Adversarial Examples via Semantic Manipulation
• A Survey on Transferability of Adversarial Examples across Deep Neural Networks