Machine Learning Algorithms In User Authentication Schemes

TL;DR

This paper reviews how various machine learning algorithms are used in user authentication schemes based on touch dynamics and device movement, highlighting current practices, benefits, limitations, and future directions.

Contribution

It provides a comprehensive overview of machine learning techniques in user authentication, focusing on touch and movement data, and discusses their advantages and challenges.

Findings

Machine learning algorithms improve user authentication accuracy.

Limitations include data variability and security concerns.

Future work suggests integrating multi-modal data and enhancing robustness.

Abstract

In the past two decades, the number of mobile products being created by companies has grown exponentially. However, although these devices are constantly being upgraded with the newest features, the security measures used to protect these devices has stayed relatively the same over the past two decades. The vast difference in growth patterns between devices and their security is opening up the risk for more and more devices to easily become infiltrated by nefarious users. Working off of previous work in the field, this study looks at the different Machine Learning algorithms used in user authentication schemes involving touch dynamics and device movement. This study aims to give a comprehensive overview of the current uses of different machine learning algorithms that are frequently used in user authentication schemas involving touch dynamics and device movement. The benefits, limitations, and suggestions for future work will be thoroughly discussed throughout this paper.