Generative Human Trajectory Recovery via Embedding-Space Conditional Diffusion

Kaijun Liu, Sijie Ruan, Liang Zhang, Cheng Long, Shuliang Wang, liang yu

TL;DR: This paper presents a novel embedding-space conditional diffusion based method for recovering human trajectories from incomplete data, outperforming existing approaches by an average of 11% in recall.

Abstract: Recovering human trajectories from incomplete or missing data is crucial for many mobility-based urban applications, e.g., urban planning, transportation, and location-based services. Existing methods mainly rely on recurrent neural networks or attention mechanisms. Though promising, they encounter limitations in capturing complex spatial-temporal dependencies in low-sampling trajectories. Recently, diffusion models show potential in content generation. However, most of proposed methods are used to generate contents in continuous numerical representations, which cannot be directly adapted to the human location trajectory recovery. In this paper, we introduce a conditional diffusion-based trajectory recovery method, namely, DiffMove. It first transforms locations in trajectories into the embedding space, in which the embedding denoising is performed, and then missing locations are recovered by an embedding decoder. DiffMove not only improves accuracy by introducing high-quality generative methods in the trajectory recovery, but also carefully models the transition, periodicity, and temporal patterns in human mobility. Extensive experiments based on two representative real-world mobility datasets are conducted, and the results show significant improvements (an average of 11% in recall) over the best baselines.

Lay Summary: People increasingly rely on location-based services—like maps, ride-sharing, and recommendation apps—but often stop sharing their location due to privacy or battery concerns. This creates “gaps” in the data that make it hard for apps and planners to understand where people actually go. We developed a new AI method that treats these gaps like a puzzle, using a generative process by conditional diffusion model to fill in missing stops in someone’s daily route. Instead of predicting a single determined path, our approach can suggest some plausible locations by learning the spatial temporal features and obtain better estimates of locations a person might have traveled, reflecting real-life variability. By recovering those hidden steps, our system can power more accurate transit planning, personalized recommendations, and better traffic management—while still respecting users’ choice to share only part of their journey.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Deep Learning->Everything Else

Keywords: Trajectory recovery, Diffusion model, Self-supervised learning, Human mobility

Submission Number: 9495