Adversarial sample generation method based on depth product quantization

Abstract

The invention provides an adversarial sample generation method based on depth product quantization. The adversarial sample generation method comprises the steps of: inputting an original image into apreset network model, so as to output a quantitative distribution center vector corresponding to the original image through the preset network model; inputting an initial adversarial sample corresponding to the original image into the preset network model to obtain an adversarial feature vector corresponding to the initial adversarial sample; based on the quantitative distribution center vector and the adversarial feature vector, determining a loss function corresponding to the initial adversarial sample; and performing back propagation on the preset network model based on the loss function toobtain an adversarial sample corresponding to the original image. According to the adversarial sample generation method, the derivable loss function is determined on the basis of the quantitative distribution center vector and the adversarial feature vector, so that the mobility and effectiveness of the adversarial sample are improved, and a basis is provided for further researching the robustness of the neural network.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to an adversarial example generation method based on depth product quantization. Background technique [0002] Achieving large-scale and high-dimensional image retrieval depends on effective image feature representation and matching. There are two main types of existing mainstream methods, mainly binary representation methods based on hash, which are obtained by fast matching speed. Wide range of applications. Another large category of methods is the data compression method dominated by product quantization, which generally has higher retrieval performance than the hash method. At present, the method of hash or product quantization based on convolutional neural network CNN makes full use of the powerful feature extraction ability of CNN, performs hash representation or product quantization on the feature layer of CNN, and jointly trains CNN parameters and ha...

AI Technical Summary

Problems solved by technology

Although there are already design methods for adversarial examples such as classification, the quantization process based on the deep product quantitative retrieval system is discrete and non-differentiable. How to effectively use the backpropagation of neural networks to obtain adversarial examples is a major difficulty.

Images