Enhancing Continuous Authentication through Touch Stroke Analysis with Explainable AI (xAI)

Muhammad Nadzmi Mohd Nizam, Ooi Shih Yin, Soodamani Ramalingam, Pang Ying Han

Research output: Contribution to conferencePaperpeer-review

Abstract

Since mobile devices are now utilized for many essential tasks, such as online banking and shopping, which call for safe authentication techniques, mobile authentication has grown in importance. While conventional techniques such as PINs and swipe patterns are widely employed, they are easily observed or guessed at, making them susceptible to social engineering attacks. Biometric authentication techniques including touch-stroke dynamics, facial recognition, and fingerprint recognition have been presented as ways to get around these restrictions. Touch-stroke dynamics is one of the most promising of them since it compliments PIN typing and swiping patterns while staying nearly transparent to users. Touch-stroke dynamics, which considers variables like speed and pressure used during swipes or taps, records the distinct patterns of how a user interacts with the screen of a mobile device. To improve continuous authentication systems, this work explores the integration of touch stroke analysis and Explainable AI (xAI). Conventional authentication techniques frequently call for user intervention at specific intervals, which can be less secure and intrusive. On the other hand, continuous authentication functions invisibly and offers a dynamic security measure by utilizing the distinct touch dynamics of users. The technology creates a comprehensive profile of user interaction behaviors by aggregating touch stroke data from multiple sessions and settings. This data is analyzed using machine learning methods such as Random Trees, Support Vector Machines (SVM), and k-Nearest Neighbors (k-NN). According to the study, the Random Tree classifier outperforms the k-NN classifier, which has an accuracy of 57.14%, with a classification accuracy of 97.08%. By implementing xAI principles, the system's decision-making process is transparent, promoting user confidence and adherence to legal requirements. The results show how touch stroke dynamics and xAI may be combined to produce a reliable, transparent, and easy-to-use continuous authentication system that will ultimately improve digital security.
Original languageEnglish
Number of pages18
Publication statusPublished - 15 Jul 2024
EventCITIC 2024: The 4th International Conference on Computer, Information Technology and Intelligent Computing - Virtual, Malaysia
Duration: 23 Jul 202425 Jul 2024
Conference number: 1571029145
https://difcon.mmu.edu.my/citic.html#1#aboutcitic

Conference

ConferenceCITIC 2024: The 4th International Conference on Computer, Information Technology and Intelligent Computing
Abbreviated titleCITIC 2024
Country/TerritoryMalaysia
Period23/07/2425/07/24
Internet address

Fingerprint

Dive into the research topics of 'Enhancing Continuous Authentication through Touch Stroke Analysis with Explainable AI (xAI)'. Together they form a unique fingerprint.

Cite this