Fast-UAP: An Algorithm for Speeding up Universal Adversarial Perturbation Generation with Orientation of Perturbation Vectors

TL;DR

This paper introduces Fast-UAP, an optimized algorithm for generating universal adversarial perturbations more efficiently by leveraging the orientation of perturbation vectors, resulting in faster generation and higher fooling rates.

Contribution

The paper proposes a novel method that improves universal perturbation generation speed and effectiveness by aggregating perturbations with similar orientations, outperforming existing UAP algorithms.

Findings

Faster universal perturbation generation compared to UAP.

Achieved a 9% higher fooling rate on average.

Requires fewer training images for effective perturbation.

Abstract

Convolutional neural networks (CNN) have become one of the most popular machine learning tools and are being applied in various tasks, however, CNN models are vulnerable to universal perturbations, which are usually human-imperceptible but can cause natural images to be misclassified with high probability. One of the state-of-the-art algorithms to generate universal perturbations is known as UAP. UAP only aggregates the minimal perturbations in every iteration, which will lead to generated universal perturbation whose magnitude cannot rise up efficiently and cause a slow generation. In this paper, we proposed an optimized algorithm to improve the performance of crafting universal perturbations based on orientation of perturbation vectors. At each iteration, instead of choosing minimal perturbation vector with respect to each image, we aggregate the current instance of universal perturbation with the perturbation which has similar orientation to the former so that the magnitude of the aggregation will rise up as large as possible at every iteration. The experiment results show that we get universal perturbations in a shorter time and with a smaller number of training images. Furthermore, we observe in experiments that universal perturbations generated by our proposed algorithm have an average increment of fooling rate by 9% in white-box attacks and black-box attacks comparing with universal perturbations generated by UAP.