Optimizing Mouse Dynamics for User Authentication by Machine Learning: Addressing Data Sufficiency, Accuracy-Practicality Trade-off, and Model Performance Challenges
Yi Wang, Chengyv Wu, Yang Liao, Maowei You
TL;DR
This work tackles the challenges of mouse-dynamics authentication by introducing a data-sufficiency framework based on Gaussian Kernel Density Estimation and KL divergence to determine when additional data provides diminishing returns. It then defines the Mouse Authentication Unit (MAU) and uses Approximate Entropy to balance segmentation length, achieving efficient yet informative behavioral representations. The Local-Time Mouse Authentication (LTMouseAuthen) framework fuses 1D-ResNet-based local feature extraction with GRU-based temporal modeling, yielding state-of-the-art performance on Balabit and DFL datasets (e.g., AUC up to 98.52% and 97.73% respectively, with low EERs) and demonstrating robustness against advanced imitation attacks. The results indicate practical viability for real-world deployment, offering data-efficient training, scalable inference, and strong defense against adversaries, while outlining directions for cross-platform adaptation and privacy-preserving implementations.
Abstract
User authentication is essential to ensure secure access to computer systems, yet traditional methods face limitations in usability, cost, and security. Mouse dynamics authentication, based on the analysis of users' natural interaction behaviors with mouse devices, offers a cost-effective, non-intrusive, and adaptable solution. However, challenges remain in determining the optimal data volume, balancing accuracy and practicality, and effectively capturing temporal behavioral patterns. In this study, we propose a statistical method using Gaussian kernel density estimate (KDE) and Kullback-Leibler (KL) divergence to estimate the sufficient data volume for training authentication models. We introduce the Mouse Authentication Unit (MAU), leveraging Approximate Entropy (ApEn) to optimize segment length for efficient and accurate behavioral representation. Furthermore, we design the Local-Time Mouse Authentication (LT-AMouse) framework, integrating 1D-ResNet for local feature extraction and GRU for modeling long-term temporal dependencies. Taking the Balabit and DFL datasets as examples, we significantly reduced the data scale, particularly by a factor of 10 for the DFL dataset, greatly alleviating the training burden. Additionally, we determined the optimal input recognition unit length for the user authentication system on different datasets based on the slope of Approximate Entropy. Training with imbalanced samples, our model achieved a successful defense AUC 98.52% for blind attack on the DFL dataset and 94.65% on the Balabit dataset, surpassing the current sota performance.