Transferable Unsupervised Outlier Detection Framework for Human Semantic Trajectories
This addresses the problem of identifying outlier behaviors in semantic trajectories for applications like healthcare and urban planning, offering a domain-agnostic approach that is more adaptable than traditional heuristic methods.
The paper tackles outlier detection in human semantic trajectories (spatial-temporal data with textual information) by proposing the TOD4Traj framework, which integrates multi-modal data and uses contrastive learning to identify regular patterns, achieving superior performance over existing models across various datasets.
Semantic trajectories, which enrich spatial-temporal data with textual information such as trip purposes or location activities, are key for identifying outlier behaviors critical to healthcare, social security, and urban planning. Traditional outlier detection relies on heuristic rules, which requires domain knowledge and limits its ability to identify unseen outliers. Besides, there lacks a comprehensive approach that can jointly consider multi-modal data across spatial, temporal, and textual dimensions. Addressing the need for a domain-agnostic model, we propose the Transferable Outlier Detection for Human Semantic Trajectories (TOD4Traj) framework.TOD4Traj first introduces a modality feature unification module to align diverse data feature representations, enabling the integration of multi-modal information and enhancing transferability across different datasets. A contrastive learning module is further pro-posed for identifying regular mobility patterns both temporally and across populations, allowing for a joint detection of outliers based on individual consistency and group majority patterns. Our experimental results have shown TOD4Traj's superior performance over existing models, demonstrating its effectiveness and adaptability in detecting human trajectory outliers across various datasets.