Toggle Main Menu Toggle Search

Open Access padlockePrints

The Newcastle University research output collection, currently available on ePrints, will shortly be moving to a new open repository platform, Figshare. To prepare for the data migration we have paused adding new content to ePrints, and will resume once the new repository is launched. During this time you will continue to have access to ePrints (but no new content will appear). We will share updates here when available.

NUSA: Drift-Aware Multimodal Authentication for Mobile Social Media

Lookup NU author(s): Dr Lei ShiORCiD

Downloads

Licence

This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Abstract

This paper investigates whether natural mobile interaction patterns exhibit stable behavioural signatures for identity verification, addressing a critical gap in real-world behavioural data mining research. Using NUSA, a custom short-form video app, we collected multimodal interaction data from 10 participants across 665 unconstrained sessions over 16 days. We propose a 50-dimensional embedding encapsulating three core dimensions: motor coordination, contextual rhythm, engagement dynamics, and integrate a drift detection mechanism defined by 2σ deviation from user-specific baselines. Cross-validation yields a mean AUC of 0.91 ± 0.03 and demonstrates robust user separability, with inter-user distances 10× greater than intra-user distances. Notably, NUSA enables lightweight on-device authentication with an inference latency of ≤ 10 ms; this design ensures full transparency, eliminates disruptive user interaction interruptions, and mitigates key limitations inherent to black-box behavioural biometric systems. Our findings confirm that behavioural rhythm mined from natural interactions remains stable across contextual shifts, thereby establishing a critical bridge between behavioural data mining, mobile HCI, and system security domains.

Publication metadata

Author(s): Jalilzade E, Shi L

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Published

Conference Name: The 30th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) special session on Data Science: Foundations and Applications

Year of Conference: 2026

Pages: 320–332

Online publication date: 25/06/2026

Acceptance date: 17/02/2026

Date deposited: 10/03/2026

Publisher: Springer

URL: https://doi.org/10.1007/978-981-92-1947-6_26

DOI: 10.1007/978-981-92-1947-6_26

ePrints DOI: 10.57711/s5n3-mm87

Library holdings: Search Newcastle University Library for this item

Series Title: Lecture Notes in Computer Science (LNCS) and Lecture Notes in Artificial Intelligence (LNAI) and Pacific-Asia Conference on Knowledge Discovery and Data Mining conference series

ISBN: 9789819219469

Share