Research map / Label-efficient and robust learning / Recognition and long-tail
Region-Point Joint Representation for Effective Trajectory Similarity Learning
Hao Long, Silin Zhou, Lisi Chen, Shuo Shang
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 |
|---|---|---|---|
| add locality-aware CNN encoding (ablation) | RePo · Porto, DFD | HR@1: 0.374 -> 0.483 | Improved |
| add node2vec structural embedding (ablation) | RePo · Chengdu, DFD | HR@1: 0.273 -> 0.392 | Improved |
| add ResNet satellite visual embedding (ablation) | RePo · Chengdu, DFD | HR@1: 0.332 -> 0.392 | Improved |
| add correlation-aware graph encoding (ablation) | RePo · Chengdu, DFD | HR@1: 0.310 -> 0.392 | Improved |
| add continuity-preserving CDE encoding (ablation) | RePo · Porto, DTW | HR@1: 0.462 -> 0.532 | Improved |
| add continuity-preserving CDE encoding (ablation) | RePo · Chengdu, DTW | HR@1: 0.403 -> 0.428 | Improved |
| add locality-aware CNN encoding (ablation) | RePo · Chengdu, DTW | HR@1: 0.344 -> 0.428 | Improved |
| use region plus point-wise features instead of grid-only baselines | RePo vs SIMformer · Porto, DTW | HR@1: 0.300 -> 0.532 | Improved |
| RePo versus best baseline under DFD | RePo · Geolife, DFD | HR@1: 0.316 -> 0.325 | No clear effect |
| evaluate on larger candidate pool (200k) without retraining | RePo · Porto, DTW (trained on 10k) | HR@1: 0.300 -> 0.192 | Hurt |
| increase embedding dimension beyond 512 (ablation) | RePo · HR@50 sensitivity | HR@50 | No clear effect |
| use more than one hard negative sample (ablation) | RePo · HR@50 sensitivity | HR@50 | Hurt |
Closest papers by meaning
- KGTS: Contrastive Trajectory Similarity Learning over Prompt Knowledge Graph Embedding (AAAI 2024 · similarity 0.87)
- TrajMamba: An Efficient and Semantic-rich Vehicle Trajectory Pre-training Model (NeurIPS 2025 · similarity 0.87)
- GTR: A General, Multi-View, and Dynamic Framework for Trajectory Representation Learning (ICML 2025 · similarity 0.86)
- TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability (NeurIPS 2025 · similarity 0.86)
- UniTraj: Learning a Universal Trajectory Foundation Model from Billion-Scale Worldwide Traces (NeurIPS 2025 · similarity 0.85)
- Holistic Semantic Representation for Navigational Trajectory Generation (AAAI 2025 · similarity 0.84)
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.
