This website requires JavaScript.

TypeFormer: Transformers for Mobile Keystroke Biometrics

Giuseppe StragapedePaula Delgado-SantosRuben TolosanaRuben Vera-RodriguezRichard GuestAythami Morales
Dec 2022
摘要
The broad usage of mobile devices nowadays, the sensitiveness of theinformation contained in them, and the shortcomings of current mobile userauthentication methods are calling for novel, secure, and unobtrusive solutionsto verify the users' identity. In this article, we propose TypeFormer, a novelTransformer architecture to model free-text keystroke dynamics performed onmobile devices for the purpose of user authentication. The proposed modelconsists in Temporal and Channel Modules enclosing two Long Short-Term Memory(LSTM) recurrent layers, Gaussian Range Encoding (GRE), a multi-headSelf-Attention mechanism, and a Block-Recurrent structure. Experimenting on oneof the largest public databases to date, the Aalto mobile keystroke database,TypeFormer outperforms current state-of-the-art systems achieving Equal ErrorRate (EER) values of 3.25% using only 5 enrolment sessions of 50 keystrokeseach. In such way, we contribute to reducing the traditional performance gap ofthe challenging mobile free-text scenario with respect to its desktop andfixed-text counterparts. Additionally, we analyse the behaviour of the modelwith different experimental configurations such as the length of the keystrokesequences and the amount of enrolment sessions, showing margin for improvementwith more enrolment data. Finally, a cross-database evaluation is carried out,demonstrating the robustness of the features extracted by TypeFormer incomparison with existing approaches.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答