基于神经网络的口令属性分析方法

Analysis Method of Password Attributes Based on Neural Network

提出了一种适用于大量口令的分析方法,通过神经网络建模分析,挖掘口令的各项属性间可能存在的关系。通过对大量不同应用类型的口令进行属性赋值,进而利用MATLAB的神经网络工具完成建模、训练学习和仿真测试,从中找出潜在的有价值的属性间关系或规则,并根据这些关系由已知的属性值推断出未知属性值,或根据这些属性规则生成反映用户口令设置习惯的暴力破解字典,为口令破解和口令恢复提供一定的指向性和帮助。

This paper proposes an analyzing method which is fit for a large number of passwords. By building up a neural network model, it can analyze and dig for potential relationships between password attributes. After attribute assignment for passwords of different application types, it can utilize neural network tool in MATLAB to finish model building, training and simulation testing, in order to find out potential and valuable relationships or rules between attributes. With such relationships, it can help to deduce unknown attribute values from the already known ones, or provide some directivity help for password cracking and recovery work by generating brute force cracking dictionary, which reflects the users＇ habits of password setting.