Frequency-Tuned Universal Adversarial Attacks

Abstract

Researchers have shown that the predictions of a convolutional neural network (CNN) for an image set can be severely distorted by one single image-agnostic perturbation, or universal perturbation, usually with an empirically fixed threshold in the spatial domain to restrict its perceivability. However, by considering the human perception, we propose to adopt JND thresholds to guide the perceivability of universal adversarial perturbations. Based on this, we propose a frequency-tuned universal attack method to compute universal perturbations and show that our method can realize a good balance between perceivability and effectiveness in terms of fooling rate by adapting the perturbations to the local frequency content. Compared with existing universal adversarial attack techniques, our frequency-tuned attack method can achieve cutting-edge quantitative results. We demonstrate that our approach can significantly improve the performance of the baseline on both white-box and black-box attacks.

References in corpus (7)

• ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Models
• Fast Feature Fool: A data independent approach to universal adversarial perturbations
• Adversarial Examples for Semantic Segmentation and Object Detection
• Characterizing Adversarial Examples Based on Spatial Consistency Information for Semantic Segmentation
• Defending Against Universal Attacks Through Selective Feature Regeneration
• A Survey of Black-Box Adversarial Attacks on Computer Vision Models
• A geometry-inspired decision-based attack

Cited by in corpus (3)

• A Frequency Perspective of Adversarial Robustness
• A Perceptual Distortion Reduction Framework: Towards Generating Adversarial Examples with High Perceptual Quality and Attack Success Rate
• Towards Imperceptible Universal Attacks on Texture Recognition