ProDiff: Prototype-Guided Diffusion for Minimal Information Trajectory Imputation

2025-05-29

AI Summary

To address the challenge of severe trajectory data sparsity—where only origin and destination points are available due to device constraints—this paper proposes the first end-to-end trajectory completion framework integrating prototype learning with denoising diffusion models. The method eschews reliance on sparse intermediate locations or auxiliary information (e.g., velocity), instead modeling human mobility patterns via prototype learning and conditioning the diffusion process solely on endpoints. A jointly optimized spatiotemporal consistency loss ensures the plausibility and coherence of generated trajectories. Evaluated on the Foursquare and WuXi datasets, the approach achieves accuracy improvements of 6.28% and 2.52%, respectively, over state-of-the-art baselines. Moreover, the Pearson correlation between generated and ground-truth trajectories reaches 0.927, demonstrating superior fidelity and structural alignment. This work establishes a novel paradigm for trajectory synthesis under extreme sparsity, advancing both model expressiveness and physical plausibility in mobility modeling.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Security and Privacy: Data transparency and provenance

Abstract

Trajectory data is crucial for various applications but often suffers from incompleteness due to device limitations and diverse collection scenarios. Existing imputation methods rely on sparse trajectory or travel information, such as velocity, to infer missing points. However, these approaches assume that sparse trajectories retain essential behavioral patterns, which place significant demands on data acquisition and overlook the potential of large-scale human trajectory embeddings. To address this, we propose ProDiff, a trajectory imputation framework that uses only two endpoints as minimal information. It integrates prototype learning to embed human movement patterns and a denoising diffusion probabilistic model for robust spatiotemporal reconstruction. Joint training with a tailored loss function ensures effective imputation. ProDiff outperforms state-of-the-art methods, improving accuracy by 6.28% on FourSquare and 2.52% on WuXi. Further analysis shows a 0.927 correlation between generated and real trajectories, demonstrating the effectiveness of our approach.

Problem

• Incomplete trajectory data due to device limitations and diverse scenarios
• Existing methods rely on sparse data, assuming essential patterns are retained
• ProDiff uses two endpoints and diffusion models for accurate imputation

Innovation

• Uses two endpoints for minimal trajectory imputation
• Integrates prototype learning for movement patterns
• Employs denoising diffusion for robust reconstruction