Research map / Video / Video segmentation and tracking
TRIDENT: Cross-Domain Trajectory Spatio-Temporal Representation via Distance-Preserving Triplet Learning
Guan Yi Jhang, Jeng-Chung Lien, Yu Hui-Ching, Hsu-Chao Lai, Jiun-Long Huang
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 DAEncoder instead of Transformer encoder (ablation) | TRIDENT with DMT · T-Drive | HR@10: 0.5038 -> 0.5643 | Improved |
| use DAEncoder instead of BiLSTM (ablation) | TRIDENT with DMT · T-Drive | HR@10: 0.5168 -> 0.5643 | Improved |
| use DAEncoder instead of GRU (ablation) | TRIDENT with DMT · T-Drive | HR@10: 0.4044 -> 0.5643 | Improved |
| use DMT loss instead of Pairwise Logistic CL loss (ablation) | TRIDENT DAEncoder · T-Drive | HR@10: 0.5481 -> 0.5643 | Improved |
| use DMT loss instead of Siamese Contrastive Loss (ablation) | TRIDENT DAEncoder · T-Drive | HR@10: 0.2201 -> 0.5643 | Improved |
| use DMT loss instead of RBF-Triplet Loss (ablation) | TRIDENT DAEncoder · T-Drive | HR@10: 0.5094 -> 0.5643 | Improved |
| add NTAP attention pooling (ablation) | TRIDENT · Badminton | HR@10: 0.1477 -> 0.1978 | Improved |
| add NTAP attention pooling (ablation) | TRIDENT · Badminton | R@10: 0.4014 -> 0.5047 | Improved |
| add NTAP attention pooling (ablation) | TRIDENT · T-Drive | HR@10: 0.5540 -> 0.5643 | Improved |
| use NTAP instead of standard attention pooling (ablation) | TRIDENT · Badminton | HR@10: 0.1444 -> 0.1978 | Improved |
| add standard attention pooling (Std-AP) (ablation) | TRIDENT · Badminton | HR@10: 0.1477 -> 0.1444 | Hurt |
| remove 50% of training data | TRIDENT · T-Drive | HR@10: 0.5643 -> 0.5195 | Hurt |
Closest papers by meaning
- GTR: A General, Multi-View, and Dynamic Framework for Trajectory Representation Learning (ICML 2025 · similarity 0.87)
- KGTS: Contrastive Trajectory Similarity Learning over Prompt Knowledge Graph Embedding (AAAI 2024 · similarity 0.85)
- UniTR: A Unified Framework for Joint Representation Learning of Trajectories and Road Networks (AAAI 2025 · similarity 0.84)
- Learned Trajectory Embedding for Subspace Clustering (CVPR 2024 · similarity 0.84)
- Generative Human Trajectory Recovery via Embedding-Space Conditional Diffusion (ICML 2025 · similarity 0.84)
- Scalable Trajectory-User Linking with Dual-Stream Representation Networks (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.
