基于击键动力学的身份认证技术的研究
首发时间:2017-11-27
摘要:传统的基于击键行为的身份认证技术主要基于时间特征,例如通过获取不同按键的按压时间和释放时间来获取一个按键的按压时间或者不同按键之间的时间间隔,通过这种方式获取的特征种类过于单一,对用户行为模式的识别效果较差。鉴于此,本文选用了非传统的数据特征,例如部分时间特征、编辑特征等,在一定程度上解决了上述问题。在分类模型的设计上,本文设计了基于最小距离的单分类模型,得到了较好的实验结果。同时本文采用基于自由文本的检测方式,直接使用用户日常的击键行为数据进行模型训练,较好地契合了现实使用场景、提高了用户体验度并可实现对用户身份的持续验证。
For information in English, please click here
Research on identity authentication technology based on keystroke dynamics
Abstract:Traditional identity authentication technology based on keystroke behavior is mainly based on the time features, for example by getting different button pressing time and releasing time for press of a button or the time intervals between different keys, the species of the characteristics obtained in this way can be too single, then the user behavior pattern recognition effect is poorer. In view of this, this paper chooses the data characteristics of the non-traditional (characteristic to distinguish with the traditional time feature), for example, partial time features (features derived from the time features), editing features, etc. To a certain extent, solve the above problem.In the design of classification model.This paper designs a single classification model based on the minimum distance, and obtains better experimental results .In this paper, the detection method based on free text is used at the same time, which directly uses the daily user keystroke behavior data for training model, throgh this way, fit the practical usage scenarios better, improve the user experience degrees and can realize the continuous verification of user identities.
Keywords: keystroke identification free text detection non-traditional feature
引用
No.****
同行评议
共计0人参与
勘误表
基于击键动力学的身份认证技术的研究
评论
全部评论0/1000