Password cracking method and system of generative adversarial network based on strategy gradient

Abstract

The invention discloses a password cracking method and system of a generative adversarial network based on strategy gradient, and the method comprises the steps of carrying out the preprocessing of a leaked password set, obtaining a training password set, and converting the training password set into a character array composed of password characters; counting character frequencies and sorting to obtain a character dictionary and a reverse index dictionary; converting the character array into a digital password array according to the character dictionary to obtain a training data set; constructing a neural network, wherein the neural network comprises a generator structure class, a roll-out structure class, a Highway structure class, a discriminator structure class and a strategy gradient algorithm; generating examples of a generator and a discriminator, and initializing parameters of the generator and the discriminator and an optimizer; pre-training the generator and the discriminator; performing adversarial training on the generator and the discriminator; and performing forward propagation by using the trained generator parameters to generate a blasting password set. According to the invention, the original GAN training is more stable, and the text with higher quality is generated.

Description

technical field [0001] The invention relates to the technical field of password cracking, in particular to a password cracking method and system based on a strategy gradient-based generative confrontation network. Background technique [0002] The development of the Internet has given birth to various application systems, such as various APPs and websites. Users use these application systems to basically use plaintext passwords for user authentication. According to the leaked password sets of some websites, it can be found that the plaintext passwords set by users have certain Regularity, such as the combination of name pinyin, birthday, phone number, etc. Existing plaintext password cracking methods include brute force cracking, dictionary cracking and neural network-based cracking methods. Brute force cracking takes a lot of time and resources; dictionary cracking is based on the statistical characteristics of plaintext passwords, compared with brute force cracking, it sa...

AI Technical Summary

Problems solved by technology

Brute force cracking takes a lot of time and resources; dictionary cracking is based on the statistical characteristics of plaintext passwords, compared with brute force cracking, it saves a lot of time and resources, but it just keeps trying the permutations and combinations of dictionaries without real password cracking. The mode of password set by the user can be learned accurately; the cracking method based on the neural network is to use the neural network to learn the distribution characteristics of the password. The existing cracking method based on RNNs and the PassGAN cracking method can only be used after the complete password is generated. To judge the password, the action of generating the password cannot be judged during the password generation process, that is to say, these two methods cannot realize the whole process of password generation guidance; experiments show that the hit rate based on the neural network cracking method is higher than that of dictionary cracking

Images