Smartphone User Authentication Based on Holding Position and Touch-Typing Biometrics

Abstract

In this advanced age, when smart phones are the norm, people utilize social networking, online shopping, and even private information storage through smart phones. As a result, identity authentication has become the most critical security activity in this period of the intelligent craze. By analyzing the shortcomings of the existing authentication methods, this paper proposes an identity authentication method based on the behavior of smartphone users. Firstly, the sensor data and touch-screen data of the smart phone users are collected through android programming. Secondly, the eigenvalues of this data are extracted and sent to the server. Thirdly, the Support Vector Machine (SVM) and Recurrent Neural Network (RNN) are introduced to train the collected data on the server end, and the results are finally yielded by the weighted average. The results show that the method this paper proposes has great FRR (False Reject Rate) and FAR (False Accept Rate).

Modeling of On-Chip Biosensor for the in Vivo Diagnosis of Hypertension in Wireless Body Area Networks

“…This sensor can further be implemented in the Wireless Body area network (WBAN), and can be integrated into Ehealthcare platforms, owing to the advancements in digitization. Some more related literature can be studied in the references [54][55][56][57][58].…”

Applications of Recurrent Neural Network for Biometric Authentication & Anomaly Detection

“…An increasingly popular form of biometric authentication is through the recognition of mouse movements or keyboard-based behavioral patterns. Rapid User Mouse Behavior Authentication (RUMBA) [22] is a novel attempt to detect patterns in mouse movements using RNNs and the architecture of this model is represented in Figure 2. The researchers took this approach because monitoring physical characteristics requires access to extra hardware like specialized sensors.…”

Deep Learning and Machine Learning, Better Together Than Apart: A Review on Biometrics Mobile Authentication

“…Their model was able to achieve 9.2% EER in mobile authentication with a triplet loss function when balancing enrollment data and performance. In [53], weighted results from and RNN and a SVM were used to authenticate keystroke data. A small dataset consisting of motion sensor, touchscreen, and temporal data from 10 users was used to train the models.…”