Research map / Probabilistic and scientific modeling / Molecules and proteins
SE(3)-Equivariant Flow Matching with Gaussian Process Priors for Geometric Trajectory Prediction
Xuyang Wang, Xinzhe Zhou, Xiaoming Duan, Jianping He
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 GP-based prior instead of standard Gaussian N(0, I) (ablation) | GP-EquiFlow · Gravity | ADE: 0.168 -> 0.105 | Improved |
| use GP-based prior instead of standard Gaussian N(0, I) (ablation) | GP-EquiFlow · Spring | ADE: 0.0286 -> 0.0019 | Improved |
| use GP-based prior instead of standard Gaussian N(0, I) (ablation) | GP-EquiFlow · Uracil | ADE: 0.108 -> 0.065 | Improved |
| use GP-based prior instead of Gaussian random walk prior (ablation) | GP-EquiFlow · Spring | ADE: 0.0058 -> 0.0019 | Improved |
| use GP-based prior instead of Gaussian random walk prior (ablation) | GP-EquiFlow · Uracil | ADE: 0.098 -> 0.065 | Improved |
| use GP-based prior instead of GeoFlow prior (ETGN-based) (ablation) | GP-EquiFlow · Gravity | ADE: 0.135 -> 0.105 | Improved |
| condition prior on last observed position instead of N(0, I) (ablation) | flow matching prior · Gravity | ADE: 0.168 -> 0.148 | Improved |
| use fixed center-of-mass prior instead of last-position prior (ablation) | flow matching prior · Gravity | ADE: 0.148 -> 0.242 | Hurt |
| increase Euler sampling steps from 10 to 100 (ablation) | GP-EquiFlow · Gravity | ADE: 0.105 -> 0.129 | Hurt |
| increase Euler sampling steps from 10 to 100 (ablation) | GP-EquiFlow · Benzene | ADE: 0.020 -> - | Hurt |
| increase Euler sampling steps from 5 to 10 (ablation) | GP-EquiFlow · Spring | ADE: 0.0033 -> 0.0019 | Improved |
| replace GeoTDM with GP-EquiFlow | probabilistic trajectory model · Gravity (N-body) | ADE: 0.256 -> 0.105 | Improved |
Closest papers by meaning
- Geometric Trajectory Diffusion Models (NeurIPS 2024 · similarity 0.88)
- Metric Flow Matching for Smooth Interpolations on the Data Manifold (NeurIPS 2024 · similarity 0.87)
- Integration-free Kernels for Equivariant Gaussian Process Modelling (ICML 2025 · similarity 0.86)
- Not-So-Optimal Transport Flows for 3D Point Cloud Generation (ICLR 2025 · similarity 0.86)
- First-Order Trajectory Matching: Fast Ensemble Predictions of Chaotic, Turbulent, Stochastic Systems (NeurIPS 2026 · similarity 0.86)
- Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling (NeurIPS 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.
