Research map / Probabilistic and scientific modeling / Molecules and proteins
Generative Active Learning for Long-tail Trajectory Prediction via Controllable Diffusion Model
Daehee Park, Monu Surana, Pranav Desai, Ashish Mehta, Reuben MV John, Kuk-Jin Yoon
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 real guidance (ablation) | GALTraj (QCNet) · WOMD | FPR5: 0.34 -> 0.22 | Improved |
| add real guidance (ablation) | GALTraj (QCNet) · WOMD | VaR999: 7.56 -> 6.05 | Improved |
| add gradient guidance (ablation) | GALTraj (QCNet) · WOMD | minFDE6: 0.601 -> 0.558 | Improved |
| add gradient guidance (ablation) | GALTraj (QCNet) · WOMD | FPR5: 0.26 -> 0.22 | Improved |
| add sampling weight decay and random time shift to real+gradient guidance (ablation) | GALTraj (QCNet) · WOMD | FPR5: 0.28 -> 0.22 | Improved |
| add sampling weight decay and random time shift to real+gradient guidance (ablation) | GALTraj (QCNet) · WOMD | minFDE6: 0.604 -> 0.558 | Improved |
| add unguided generative augmentation (naive concatenation) (ablation) | QCNet vanilla · WOMD | FPR5: 0.42 -> 0.38 | Improved |
| replace naive generation with tail-aware generation (ablation) | QCNet · WOMD | Top 1% minADE6: 4.56 -> 3.43 | Improved |
| resampling tail samples instead of vanilla training | QCNet · WOMD | minFDE6: 0.654 -> 0.668 | Hurt |
| GALTraj training instead of vanilla | MTR · WOMD | FPR5: 0.99 -> 0.65 | Improved |
| GALTraj training instead of vanilla | QCNet · AV2 | FPR5: 0.35 -> 0.19 | Improved |
| GALTraj training instead of vanilla on training split | QCNet · WOMD train split | Top 1%: 7.38 -> 2.29 | Improved |
Closest papers by meaning
- ActiveAD: Planning-Oriented Active Learning for End-to-End Autonomous Driving (CVPR 2026 · similarity 0.88)
- NOMAD: Lifelong Trajectory Planning via Non-Parametric Bayesian Memory-Adaptive Diffusion Experts (ICML 2026 · similarity 0.88)
- Solving Motion Planning Tasks with a Scalable Generative Model (ECCV 2024 · similarity 0.87)
- STraj: Self-training for Bridging the Cross-Geography Gap in Trajectory Prediction (AAAI 2025 · similarity 0.86)
- LaKD: Length-agnostic Knowledge Distillation for Trajectory Prediction with Any Length Observations (NeurIPS 2024 · similarity 0.86)
- Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous and Instruction-guided Driving (CVPR 2024 · 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.
