Black box attack confrontation sample generation method and system based on Gaussian homotopy optimization

CN118940258AActive Publication Date: 2024-11-12HUAZHONG UNIV OF SCI & TECH

Patent Information

Application Number
CN202410977215.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-04-29
Filing Date
2024-07-22
Publication Date
2024-11-12
Estimated Expiration
2044-07-22

Smart Images

  • Figure CN118940258A_ABST
    Figure CN118940258A_ABST
Patent Text Reader

Abstract

The application requires priority, the application number of the previous application is 2024105299174, the name is a black box attack confrontation sample generation method and system based on Gaussian homotopy optimization, and the priority day is April 29, 2024. The invention discloses a black box attack confrontation sample generation method and system based on Gaussian homotopy optimization, and belongs to the field of deep learning image classification model reliability verification, and the method comprises the steps: constructing a smoothing factor adjustment model, and initializing the smoothing factor adjustment model and a continuous path learning model; the smoothing factor adjusting model outputs a smoothing factor s; the continuous path learning model outputs sample disturbance x; gaussian smoothing processing is carried out on an original adversarial sample generation target function by using a smoothing factor s to obtain a plurality of Gaussian homotopy functions G (x, s), and the adversarial sample is efficiently generated by jointly training a smoothing factor adjustment model and a continuous path learning model by taking minimization of a mean value of the plurality of Gaussian homotopy functions G (x, s) as a target.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Black box attack confrontation sample generation method and system

    CN113704758A

  • Multi-agent collaborative confrontation decision-making method and device based on reinforcement learning

    CN117273057A

  • Learning sparsity-constrained gaussian graphical models in anomaly detection

    US20200057956A1

Cited By

  • Black box antagonism rate-distortion attack method and system based on feature proxy and medium

    CN120281918A