Relative Attention-based One-Class Adversarial Autoencoder for Continuous Authentication of Smartphone Users

TL;DR

This work tackles continuous smartphone authentication by eliminating the need for impostor data and dual-model training. It introduces a relative attention-based one-class adversarial autoencoder that uses a latent space discriminator and a sample discriminator, augmented by a convolutional projection-based relative attention layer to capture rich contextual behavioral patterns from multimodal sensor data. The approach achieves strong results across HMOG, BrainRun, and IDNet, with $EER$ around 1% and AUROC near 0.997–0.998, at a swift $0.7\mathrm{s}$ decision interval, and shows robustness under random and impersonation attacks while maintaining modest mobile overhead. Ablation studies confirm the importance of the relative attention layer and two discriminators in achieving these gains. Overall, the method offers a practical, scalable solution for real-time, privacy-preserving continuous authentication on resource-constrained devices.

Abstract

Behavioral biometrics-based continuous authentication is a promising authentication scheme, which uses behavioral biometrics recorded by built-in sensors to authenticate smartphone users throughout the session. However, current continuous authentication methods suffer some limitations: 1) behavioral biometrics from impostors are needed to train continuous authentication models. Since the distribution of negative samples from diverse attackers are unknown, it is a difficult problem to solve in real-world scenarios; 2) most deep learning-based continuous authentication methods need to train two models to improve authentication performance. A deep learning model for deep feature extraction, and a machine learning-based classifier for classification; 3) weak capability of capturing users' behavioral patterns leads to poor authentication performance. To solve these issues, we propose a relative attention-based one-class adversarial autoencoder for continuous authentication of smartphone users. First, we propose a one-class adversarial autoencoder to learn latent representations of legitimate users' behavioral patterns, which is trained only with legitimate smartphone users' behavioral biometrics. Second, we present the relative attention layer to capture richer contextual semantic representation of users' behavioral patterns, which modifies the standard self-attention mechanism using convolution projection instead of linear projection to perform the attention maps. Experimental results demonstrate that we can achieve superior performance of 1.05% EER, 1.09% EER, and 1.08% EER with a high authentication frequency (0.7s) on three public datasets.

Figures (7)

• Figure 1: The architecture of relative attention-based one-class adversarial autoencoder: the denoising autoencoder with relative attention layer, the latent space discriminator, and the sample discriminator with relative attention layer.
• Figure 2: The relative attention layer, learning the dependence between a unit and its neighborhoods.
• Figure 3: Architecture of the relative attention-based one-class adversarial autoencoder. Different layers are represented with different colors.
• Figure 4: Architecture of the proposed continuous authentication system.
• Figure 5: The sensory readings of three-axis from different sensors. (a) sensory data sequence of three-axis from the accelerometer; (b) sensory data sequence of three-axis from the gyroscope; (c) sensory data sequence of three-axis from the magnetometer.