Attack method and device based on genetic algorithm and computer program product

Abstract

The invention provides an attack method and device based on a genetic algorithm, electronic equipment, a storage medium and a computer program product, relates to the technical field of artificial intelligence, in particular to computer vision, image recognition and deep learning technologies, and can be used in an attack resisting scene. According to the specific implementation scheme, a target image is acquired; starting from the initial disturbance range, determining an adversarial sample of the target model based on the target image in the current disturbance range by adopting a genetic algorithm, and increasing the disturbance range under the condition that the adversarial sample is not successfully determined so as to continuously carry out adversarial attack on the target model until a preset end condition is met; and determining attack data finally obtained based on the attack target model. According to the method and the device, the problems of high calculation cost and low confrontation sample determination efficiency caused by directly attacking the target model in a relatively large disturbance range can be effectively solved, and the attack efficiency is improved.

Description

technical field [0001] The present disclosure relates to the field of artificial intelligence technology, specifically to computer vision, image recognition and deep learning technologies, and in particular to an attack method, device, electronic device, storage medium, and computer program product based on a genetic algorithm, which can be used in adversarial attack scenarios. Background technique [0002] Image processing is an important application of artificial intelligence technology. With the continuous popularization of image processing models based on neural network models, many attack algorithms for image recognition models have emerged in recent years. By adding a small perturbation to the original image, the recognition result of the model to the changed image can be disturbed, but it does not affect the recognition of the image by the human eye. If this technology is used improperly, it will cause illegal content to bypass the automatic review model and be releas...

AI Technical Summary

Problems solved by technology

[0003] The current attack method generally generates random disturbances within the maximum permissible disturbance range to iteratively find adversarial samples, which requires a large amount of calculation and takes a long time to conduct an attack.

Images