Research map / Learning theory and algorithms / Prediction and decision losses
Towards Efficient Selection of Activity Trajectories based on Diversity and Coverage
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the full model with mean-pooling rather than LSTM | trajectory similarity learning · T-Drive | HR@50: 0.4965 -> 0.5301 | Helped |
| use the full model with mean-pooling rather than LSTM + ITA | trajectory similarity learning · NYCTL | HR@50: 0.6150 -> 0.6216 | Helped |
| add inter-trajectory attention (ITA) to LSTM (ablation) | LSTM trajectory embeddings · T-Drive | HR@50: 0.4965 -> 0.5197 | Helped |
| use the full model instead of BiLSTM without ITA (ablation) | trajectory similarity learning · T-Drive | HR@50: 0.5028 -> 0.5301 | Helped |
| use mean-pooling instead of last hidden states (ablation) | full trajectory embedding model · T-Drive | HR@50: 0.5259 -> 0.5301 | Helped |
| use mean-pooling instead of self-attention (ablation) | full trajectory embedding model · T-Drive | HR@50: 0.5131 -> 0.5301 | Helped |
| use mean-pooling instead of max-pooling (ablation) | full trajectory embedding model · NYCTL | HR@50: 0.5838 -> 0.6216 | Helped |
| use self-attention instead of mean-pooling (ablation) | full trajectory embedding model · NYCTL | R10@50: 0.8365 -> 0.8197 | Hurt |
| add upper-bound pruning (ablation) | greedy trajectory selection · large explored regions | elapsed time: up to 8.7× | Helped |
| add data partitioning to the heap-based selector (ablation) | greedy trajectory selection · T-Drive and NYCTL | elapsed time | Helped |
| use the proposed greedy selector instead of MaxMin | trajectory subset selection · T-Drive | coverage ratio: up to 7.6× | Helped |
| use mean-pooling rather than self-attention (ablation) | trajectory embedding model · T-Drive and NYCTL | similarity approximation effectiveness | Helped |
Closest papers by meaning
- Scalable Trajectory-User Linking with Dual-Stream Representation Networks (AAAI 2025 · similarity 0.83)
- TAU: Trajectory Data Augmentation with Uncertainty for Next POI Recommendation (AAAI 2024 · similarity 0.82)
- Taming the Long Tail in Human Mobility Prediction (NeurIPS 2024 · similarity 0.82)
- Towards Predicting Any Human Trajectory In Context (NeurIPS 2025 · similarity 0.81)
- Holistic Semantic Representation for Navigational Trajectory Generation (AAAI 2025 · similarity 0.81)
- Pedestrian Trajectory Prediction with Missing Data: Datasets, Imputation, and Benchmarking (NeurIPS 2024 · 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.