Research map / Probabilistic and scientific modeling / Molecules and proteins
Action-Based Contrastive Learning for Trajectory Prediction
Marah Halawa, Olaf Hellwich, Pia Bideau
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 CVAE synthetic positive/negative sampling (ablation) | ABC+ · PIE | C-FDE: 213 -> 191 | Improved |
| add CVAE synthetic positive/negative sampling (ablation) | ABC+ · JAAD | C-FDE: 425 -> 409 | Improved |
| add CVAE synthetic positive/negative sampling (ablation) | ABC+ · TITAN | C-FDE: 951 -> 843 | Improved |
| use action-based contrastive loss instead of SimCLR batch contrastive loss (ablation) | ABC+ · PIE | C-FDE: 399 -> 213 | Improved |
| use action-based contrastive loss instead of SimCLR batch contrastive loss (ablation) | ABC+ · JAAD | C-FDE: 608 -> 425 | Improved |
| use action-based contrastive loss instead of SimCLR batch contrastive loss (ablation) | ABC+ · TITAN | C-FDE: 1805 -> 951 | Improved |
| add action-based contrastive framework to BiTraP | BiTraP · PIE | C-FDE: 261 -> 191 | Improved |
| add action-based contrastive framework to BiTraP | BiTraP · JAAD | C-FDE: 565 -> 409 | Improved |
| add action-based contrastive framework to BiTraP | BiTraP · JAAD | ADE 0.5s: 38 -> 40 | No clear effect |
| add action-based contrastive framework to BiTraP | BiTraP · TITAN | C-FDE: 989 -> 843 | Improved |
| use action labels via contrastive loss instead of concatenating action to embedding (Titan-AP) | Titan-AP · TITAN (10 to 20 frames) | FDE: 55.80 -> 46.84 | Improved |
| set contrastive loss weight beta to 0.75 versus 0.25 and 0.5 (ablation) | ABC+ · PIE, JAAD, TITAN | C-FDE | Improved |
Closest papers by meaning
- TrajCLIP: Pedestrian trajectory prediction method using contrastive learning and idempotent networks (NeurIPS 2024 · similarity 0.91)
- Towards Predicting Any Human Trajectory In Context (NeurIPS 2025 · similarity 0.88)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.87)
- Pedestrian Trajectory Prediction with Missing Data: Datasets, Imputation, and Benchmarking (NeurIPS 2024 · similarity 0.86)
- Multi-modal Knowledge Distillation-based Human Trajectory Forecasting (CVPR 2025 · similarity 0.86)
- Adapting to Observation Length of Trajectory Prediction via Contrastive Learning (CVPR 2025 · similarity 0.86)
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.
