Generalization adversarial sample group generation method based on easily-confusable category feature injection

By introducing class perturbation to update momentum using easily confused category feature vectors, a generalized perturbation sample set for a single category is generated. This solves the problems of excessive modification to the original image and low efficiency in existing generalized perturbation templates, and achieves efficient and covert adversarial sample generation.

CN121145982BActive Publication Date: 2026-06-09CHINESE PEOPLES LIBERATION ARMY UNIT 32801
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
CHINESE PEOPLES LIBERATION ARMY UNIT 32801
Filing Date
2025-09-05
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing general adversarial attack methods cannot generate perturbation samples for single-class images. Generalized perturbation templates modify the original image too much, lack sufficient concealment, and have low generation efficiency.

Method used

In the process of generating class-oriented generalized perturbation sample sets, class perturbation update momentum in the direction of easily confused class feature vectors is introduced. By injecting easily confused class features, adversarial examples are updated to neighboring classes more quickly, generating generalized perturbation sample sets for a single class.

Benefits of technology

It achieves the generation of generalized perturbation sample groups for single-category images, reducing excessive perturbation and improving attack efficiency and stealth.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121145982B_ABST
    Figure CN121145982B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of generalization adversarial sample group generation method based on easily confused class feature injection, belong to the technical field of adversarial attack of computer vision field.It introduces the class disturbance update momentum in the direction of easily confused class feature vector in the process of generating class-oriented generalization disturbance sample group, the class disturbance update momentum makes the adjacent easily confused class of adversarial example faster update, to realize the generalization attack to class.The present application only integrates the features of easily confused class in the same class image, forms disturbance template, realizes class-oriented generalization disturbance sample group;Variable step size iterative attack means is used to explore easily confused class in the process of neural network classification, realize easily confused feature injection, while introducing class momentum parameter in the process of iterative attack, improve attack efficiency and alleviate the over disturbance shortcomings of generalization disturbance.
Need to check novelty before this filing date? Find Prior Art