M-STAR: Multi-Scale Spatiotemporal Autoregression for Human Mobility Modeling

2025-12-08

AI Summary

Existing methods for generating long-term human trajectories (e.g., weekly-scale) suffer from low efficiency and lack of explicit multi-scale modeling. This paper proposes M-STAR, a Multi-scale Spatio-Temporal Autoregressive framework, introducing— for the first time—a hierarchical spatio-temporal tokenizer and a coarse-to-fine, level-wise autoregressive generation mechanism. M-STAR integrates hierarchical encoding with a Transformer-based decoder to enable efficient, high-fidelity end-to-end trajectory modeling. Its core innovation lies in explicitly capturing mobility pattern dependencies across spatio-temporal granularities (e.g., city-level → neighborhood-level → POI-level). Experiments on two real-world datasets demonstrate that M-STAR achieves over 3.2× faster generation speed than prior approaches, while synthesized trajectories significantly outperform state-of-the-art methods across key metrics—including spatial distribution fidelity, trip frequency accuracy, and dwell-time consistency.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Modeling human mobility is vital for extensive applications such as transportation planning and epidemic modeling. With the rise of the Artificial Intelligence Generated Content (AIGC) paradigm, recent works explore synthetic trajectory generation using autoregressive and diffusion models. While these methods show promise for generating single-day trajectories, they remain limited by inefficiencies in long-term generation (e.g., weekly trajectories) and a lack of explicit spatiotemporal multi-scale modeling. This study proposes Multi-Scale Spatio-Temporal AutoRegression (M-STAR), a new framework that generates long-term trajectories through a coarse-to-fine spatiotemporal prediction process. M-STAR combines a Multi-scale Spatiotemporal Tokenizer that encodes hierarchical mobility patterns with a Transformer-based decoder for next-scale autoregressive prediction. Experiments on two real-world datasets show that M-STAR outperforms existing methods in fidelity and significantly improves generation speed. The data and codes are available at https://github.com/YuxiaoLuo0013/M-STAR.

Problem

• Generates long-term human mobility trajectories efficiently
• Models explicit multi-scale spatiotemporal patterns in mobility
• Improves fidelity and speed of synthetic trajectory generation

Innovation

• Multi-scale spatiotemporal tokenizer for hierarchical mobility patterns
• Transformer-based decoder for next-scale autoregressive prediction
• Coarse-to-fine prediction process for long-term trajectory generation