Region-Point Joint Representation for Effective Trajectory Similarity Learning

2025-11-17

AI Summary

Existing trajectory similarity learning methods struggle to adequately model multi-granularity trajectory information. To address this, we propose RePo, the first model that jointly captures region-level spatial-semantic context and point-level multimodal motion features: it employs grid-based sequences to encode macroscopic structural semantics while preserving raw GPS points to represent fine-grained motion patterns. RePo introduces an adaptive router network and a cross-attention mechanism to effectively fuse these dual-granularity representations, and incorporates contrastive loss with hard negative mining to enhance training efficacy. Extensive experiments on multiple benchmark datasets demonstrate that RePo achieves an average accuracy improvement of 22.2% over state-of-the-art methods across all evaluation metrics, significantly advancing both accuracy and generalizability in trajectory similarity computation.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Computer Vision: Motion & Tracking
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Recent learning-based methods have reduced the computational complexity of traditional trajectory similarity computation, but state-of-the-art (SOTA) methods still fail to leverage the comprehensive spectrum of trajectory information for similarity modeling. To tackle this problem, we propose extbf{RePo}, a novel method that jointly encodes extbf{Re}gion-wise and extbf{Po}int-wise features to capture both spatial context and fine-grained moving patterns. For region-wise representation, the GPS trajectories are first mapped to grid sequences, and spatial context are captured by structural features and semantic context enriched by visual features. For point-wise representation, three lightweight expert networks extract local, correlation, and continuous movement patterns from dense GPS sequences. Then, a router network adaptively fuses the learned point-wise features, which are subsequently combined with region-wise features using cross-attention to produce the final trajectory embedding. To train RePo, we adopt a contrastive loss with hard negative samples to provide similarity ranking supervision. Experiment results show that RePo achieves an average accuracy improvement of 22.2% over SOTA baselines across all evaluation metrics.

Problem

• Improving trajectory similarity computation by leveraging comprehensive trajectory information
• Jointly encoding region-wise and point-wise features for spatial context
• Capturing both spatial context and fine-grained movement patterns effectively

Innovation

• Jointly encodes region-wise and point-wise trajectory features
• Uses cross-attention to fuse spatial context with movement patterns
• Employs contrastive learning with hard negative samples