Research map / Learning theory and algorithms / Prediction and decision losses
GeoPTH: A Lightweight Approach to Category-Based Trajectory Retrieval via Geometric Prototype Trajectory Hashing
Yang Xu, Zuliang Yang, Kai Ming Ting
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Hausdorff instead of Frechet in quantization (ablation) | GeoPTH · Cyclists (L=64) | mAP: 0.878 -> 0.971 | Improved |
| use Hausdorff instead of DTW in quantization (ablation) | GeoPTH · Casia (L=64) | mAP: 0.826 -> 0.909 | Improved |
| use Hausdorff instead of Frechet in quantization (ablation) | GeoPTH · Traffic (L=64) | mAP: 0.989 -> 0.964 | Hurt |
| use Hausdorff instead of DTW in quantization (ablation) | GeoPTH · Gowalla (L=64) | mAP: 0.265 -> 0.277 | No clear effect |
| use GeoPTH instead of Hausdorff distance | GeoPTH · 7 trajectory datasets | mAP: Cyclists 0.929 -> 0.971; Gowalla 0.451 -> 0.277 | Mixed |
| use GeoPTH binary codes instead of binarized learned embeddings | GeoPTH · 7 trajectory datasets | average rank: 4.62 -> 1.57 | Improved |
| increase code length from 32 to 128 (ablation) | GeoPTH · Gowalla | mAP: 0.219 -> 0.348 | Improved |
| use GeoPTH instead of Hausdorff for execution time | GeoPTH · Gowalla (L=64) | execution time (s): 446 -> 8 | Improved |
| binarize dense learned embeddings | TrajGAT, Traj2Hash etc. · 7 trajectory datasets | mAP: TrajGAT Cyclists 0.965 -> 0.908 | Hurt |
| increase codebook size psi beyond optimum (ablation) | GeoPTH · Geolife | mAP | Mixed |
| vary prototype size k (1 to 20) (ablation) | GeoPTH · Gowalla and Geolife | mAP | No clear effect |
Closest papers by meaning
- KGTS: Contrastive Trajectory Similarity Learning over Prompt Knowledge Graph Embedding (AAAI 2024 · similarity 0.83)
- GTR: A General, Multi-View, and Dynamic Framework for Trajectory Representation Learning (ICML 2025 · similarity 0.83)
- TrajMamba: An Efficient and Semantic-rich Vehicle Trajectory Pre-training Model (NeurIPS 2025 · similarity 0.82)
- PLMTrajRec: A Scalable and Generalizable Trajectory Recovery Method with Pre-trained Language Models (NeurIPS 2025 · similarity 0.82)
- Learned Trajectory Embedding for Subspace Clustering (CVPR 2024 · similarity 0.81)
- Beyond Chamfer Distance: Granular Order-aware Evaluation Metric For Online Mapping (NeurIPS 2026 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
