Continuous User Authentication with Earable Sensors using Machine Learning

Continuous User Authentication with Earable Sensors using Machine Learning

Introduction

The increasing reliance on mobile devices and the proliferation of sensitive data necessitate robust user authentication mechanisms. Traditional methods like passwords and PINs are not possible on earable (or hearable devices). Continuous user authentication offers a more seamless and secure solution by verifying a user's identity throughout a session based on their behavioral patterns which can be captured by motion sensors (IMU: inertial measurement unit). Its efficiency in smartphones or AR devices has been shown before and it is interesting to explore a similar approach on earables.

This project explores the potential of earable Inertial Measurement Units (IMUs) for continuous user authentication. Earables are gaining popularity due to their comfortable and unobtrusive nature. IMUs embedded in these devices can capture a user's head movements and gestures, potentially serving as unique biometric identifiers.

Task

The aim is evaluating the suitability of example datasets consisting of head activities and gestures for user authentication. You are expected to implement and test machine learning algorithms to evaluate a continuous user authentication system that leverages the data from earable sensor data. The performance will be assessed in terms of accuracy, error rates (false acceptance, false rejection rates)

Multiple students are accepted.

Work

40% Theory, 40% Programming, 20% Writing

Contact

Özlem Durmaz İncel, Associate Professor, Pervasive Systems (ozlem.durmaz@utwente.nl)