Counterfeit domain name training data generation method and system based on ED-GAN

Abstract

The invention discloses a counterfeit domain name training data generation method and system based on an EDGAN. The method comprises the steps: designing an encoder and a decoder of a domain name aredesigned; combining a domain name encoder, a domain name decoder and a GAN neural network, designing a character-level domain name generative adversarial network model to generate similar counterfeitdomain name samples, and predicting and detecting counterfeit domain names; and finally, through multi-classifier parameter performance comparison, carrying out validity check on the generated counterfeit domain name sample data. In order to maximize the utilization of the characteristic that GAN can directly sample and learn samples, data is not subjected to complex processing and transformation(such as a convolution layer and a pooling layer), but is directly input into a GAN original model for learning and training, so that the real characteristics of the data can be kept; the structure ofthe domain name encoder and the domain name decoder has the characteristics of simplifying and being close to the original data, so that the real characteristics of the data can be kept to the maximum extent.

Description

technical field [0001] The invention belongs to the field of deep learning and information security, in particular to an ED-GAN-based method and system for generating counterfeit domain name training data, and belongs to counterfeit domain name protection technology. Background technique [0002] With the rapid development of Internet applications, the interests carried by the Internet are increasing, and attacks against the domain name system of network communication are also becoming more frequent, which has caused a great impact on network security. Among them, phishing domain name attack has become one of the important issues threatening the safe operation of the Internet due to its low attack cost, wide range of harm, and diversified profit means. It lures users to visit the domain name by registering a domain name similar to a legitimate domain name (such as facebo0k.com, gooqle.com, etc.) instead of the domain name of the target website to publish fake advertisements,...

AI Technical Summary

Problems solved by technology

[0011] In the detection of counterfeit domain names, in order to solve the problems of unbalanced data sets and difficult identification of new attack samples, this invention will introduce the ED-GAN character-level domain name generation model to generate usable attack data

Images