Intelligent Defense Algorithm Recommendation Method and System Based on Multiple Adversarial Example Attacks

Abstract

The invention discloses a method for recommending an intelligent defense algorithm based on various adversarial example attacks, comprising the following steps: cleaning the original data set to obtain a training set; using the training set to train a target recognition model to obtain a target attack model; selecting different The adversarial example attack algorithm and the target attack model; quantitatively evaluate the attack success rate of each adversarial example attack algorithm, and select the target attack algorithm according to the attack success rate; for each target attack algorithm, apply the adversarial attack defense algorithm one by one for defense ; According to the defense result, recommend the corresponding defense algorithm against attack to the user. The invention also discloses an intelligent defense algorithm recommendation system adopted in the intelligent defense algorithm recommendation method. For a specific target model, the intelligent defense algorithm recommendation system of the present invention can intelligently recommend an effective defense algorithm, thereby reducing the loss caused to the target model by adversarial example attacks.

Description

technical field [0001] The invention relates to the technical field of anti-example attacks and defenses in machine learning model security and privacy issues, and in particular to an intelligent defense algorithm recommendation method and system based on various anti-example attacks. Background technique [0002] Machine learning is the core of artificial intelligence. In recent years, machine learning has achieved unprecedented development, and its application has spread across various fields of artificial intelligence. Especially in the fields of data mining, computer vision, natural language processing, speech and handwriting recognition, driverless driving, etc., the application of machine learning has achieved great success. Nowadays, machine learning has received extensive attention and in-depth research, both in academia and industry. The development of machine learning has entered a new stage. Various machine learning algorithms and models emerge in an endless stre...

AI Technical Summary

Problems solved by technology

At present, most of the defense algorithms are a static defense method that only targets specific examples of attacks, and there is no unified defense algorithm recommendation and evaluation system

Images