General disturbance generation method based on generative adversarial network

Abstract

The invention discloses a general disturbance generation method based on a generative adversarial network. Firstly, the generative adversarial network generates general disturbance to obtain an adversarial sample; discriminating the adversarial sample and the original sample by a discriminating network, calculating a discriminating network objective function, and performing back propagation for optimization; and finally, predicting adversarial sample classification by a deep learning model, discriminating adversarial samples by a discriminant network, calculating and generating a network objective function, and performing back propagation for optimization. The GAN-based general disturbance generation method provided by the invention can provide a thought of machine learning model safety research for users in the fields of computer vision, deep learning and the like.

Description

Technical field [0001] The invention belongs to the technical field of deep learning, and specifically relates to a general disturbance generation method in a black box scenario based on a generation confrontation network. Background technique [0002] In 2012, in the ImageNet Large-scale Visual Recognition Challenge, Deep Neural Networks (DNNs) obtained the best image classification results at the time and began to gain widespread attention from the industry. In recent years, with the improvement of big data technology and computing performance, deep learning has developed rapidly, and more and more applications in real life have begun to use deep learning model applications. For example, autonomous driving technology uses deep learning to complete object detection, reinforcement learning, multi-modal learning, etc.; Apple uses deep learning to complete facial recognition-based biometric authentication technology; behavior-based malware detection uses deep learning to discover s...

AI Technical Summary

Problems solved by technology

The current general perturbation generation methods require the attacker to have white-box access to the model, but in real scenarios, the attacker is often limited in authority and can only access the final output value of the deep learning model

Images