Siamese Multiple Attention Temporal Convolution Networks for Human Mobility Signature Identification

TL;DR

HuMID identifies whether two GPS trajectories come from the same driver and is challenged by long trajectories and the need for local-spatial insights. The authors propose Siamese MA-TCN, a dual-branch network that fuses multi-head self-attention with depthwise separable temporal convolutions and a multi-scale aggregation attention module, augmented by a profile embedding. The model learns a dissimilarity score $D_\theta(X_1,X_2)$ and is trained with binary cross-entropy, achieving strong results on Shenzhen and Chengdu taxi datasets with improved recall and F1 and faster convergence. This approach offers scalable, efficient driver identification using GPS trajectories and has practical implications for ride-hailing safety and fraud prevention.

Abstract

The Human Mobility Signature Identification (HuMID) problem stands as a fundamental task within the realm of driving style representation, dedicated to discerning latent driving behaviors and preferences from diverse driver trajectories for driver identification. Its solutions hold significant implications across various domains (e.g., ride-hailing, insurance), wherein their application serves to safeguard users and mitigate potential fraudulent activities. Present HuMID solutions often exhibit limitations in adaptability when confronted with lengthy trajectories, consequently incurring substantial computational overhead. Furthermore, their inability to effectively extract crucial local information further impedes their performance. To address this problem, we propose a Siamese Multiple Attention Temporal Convolutional Network (Siamese MA-TCN) to capitalize on the strengths of both TCN architecture and multi-head self-attention, enabling the proficient extraction of both local and long-term dependencies. Additionally, we devise a novel attention mechanism tailored for the efficient aggregation of multi-scale representations derived from our model. Experimental evaluations conducted on two real-world taxi trajectory datasets reveal that our proposed model effectively extracts both local key information and long-term dependencies. These findings highlight the model's outstanding generalization capabilities, demonstrating its robustness and adaptability across datasets of varying sizes.

Figures (3)

• Figure 1: Siamese MA-TCN framework
• Figure 2: Multi-Head Self-Attention Double ModernTCN Block
• Figure 3: Multiple Attention Temporal Convolutional Network