Adversarial defense method of image classification network and related devices

Abstract

The invention relates to an adversarial defense method of an image classification network and a related devices and belongs to the technical field of image classification, and the method comprises thesteps: inputting an original image sample and an adversarial attack sample into a deep neural network, so as to extract the input features of a target layer, higher than a preset number of layers, ofthe deep neural network; generating a loss function of the deep neural network according to the input features to serve as an adversarial defense denoising device; utilizing the adversarial defense denoiser to denoise the adversarial attack sample to obtain a denoised adversarial attack sample; regularizing the loss function of the deep neural network to obtain a regularized deep neural network;and inputting the original image sample and the denoised countermeasure attack sample into the regularized deep neural network to obtain a classification result of the original image. According to thescheme, the defense capability of the image classification deep neural network can be effectively improved.

Description

technical field [0001] The present application relates to the technical field of image classification, in particular, to an image classification network confrontation defense method and related devices. Background technique [0002] With the in-depth application in the fields of image, voice and video, the requirements for information security of deep neural network are getting higher and higher. Although deep neural network can have very high accuracy in the process of image classification, However, adding slight noise perturbation to the samples will often cause the classification error of the neural network. Due to its vulnerability to adversarial samples, it is required that deep neural networks improve their defense capabilities and reduce the possibility of adversarial samples deceiving the network. [0003] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the ap...

AI Technical Summary

Problems solved by technology

[0002] With the in-depth application in the fields of image, voice and video, the requirements for information security of deep neural network are getting higher and higher. Although deep neural network can have very high accuracy in the process of image classification, However, adding slight noise perturbation to the samples will often cause the classification error of the neural network.

Images