Offline Password Set Generation Scheme Based On Recurrent Neural Network

With the development of Internet technology,there are various user authentication methods,such as face recognition and iris recognition.However,password authentication is still a mainstream way of network user identity authentication because of its ease of use.Certification itself has many flaws.In previous studies,password authentication has been highly valued by researchers involved in computer security,and some solutions for cracking and evaluating password security have been proposed.This paper analyzes the leaked password set on the network firstly,and then combines the recurrent neural network and natural language processing technology to proposed two offline password generation methods.The password generation model proposed in this paper is mainly based on the recurrent neural network,including the redesign and improvement of network structure to better meet the needs of password research.Specific research content includes:· The characteristics of the user's password distribution are studied by multi-dimensional analysis of the collected eight types of user password sets.It includes statistics on password frequency to find out which passwords are frequently used by users;statistics on the distribution of password text structure to study the habit of setting passwords;and length distribution analysis to study user preferences for password length selection.After that,we analyze the behavior of the user's reuse of the password and the effect of the personal information on the user's password setting by using a password set containing personal information.· As a result of the analysis of the password set,we found that different user groupshave a certain degree of similarity when setting passwords.Based on this observation phenomenon,we propose a neural network password generation model based on group attributes.The generation model takes into account the habit of setting passwords for different user groups,and can generate a password set for a specific user in a targeted manner.At the same time,we also propose a strength evaluation method based on password probability based on the model.· Since the traditional character-level password generation model does not take into account the user's language habits,it generates some meaningless character sequences,which reduces the efficiency of password guessing.In this paper,the model is improved.Firstly,a password segmentation algorithm based on hidden Markov model is used to decompose the password into tokens.Then the token is used as the basic unit of the password,and the word vector technique is used to design and generate the model.Ability to generate a more realistic set of passwords.Finally,based on hidden Markov model and recurrent neural network model,an attack strategy based on password probability is constructed,which improves the efficiency of password guessing to some extent.