Research map / Image generation / Fast diffusion sampling
Flow Expansion via Verifier-Constrained Noised State Space Exploration
Riccardo De Santi, Kimon Protopapas, Ya-Ping Hsieh, Andreas Krause
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 |
|---|---|---|---|
| replace S-MEME with G-FE (add verifier constraint) | S-MEME · Global expansion toy | validity: 0.73 -> 0.99 | Improved |
| replace S-MEME with G-FE | S-MEME · Global expansion toy | entropy: 2.17 -> 1.97 | Hurt |
| replace CONSTR with G-FE (add expansion) | CONSTR · Global expansion toy | entropy: 0.72 -> 1.97 | Improved |
| replace FDC with L-FE (add weak verifier) | FDC · Local expansion toy | validity: 0.74 -> 0.89 | Improved |
| replace pre-trained model with L-FE | pre-trained FlowMol · QM9 conformers | VENDI diversity: 89 -> 100 | Improved |
| replace FDC with L-FE | FDC · QM9 conformers | validity (%): 69% -> 81% | Improved |
| replace pre-trained model with L-FE | pre-trained FlowMol · GEOM-Drugs conformers | diversity and validity: 476 -> 529; 72% -> 82% | Improved |
| replace FDC with L-FE | FDC · GEOM-Drugs conformers | validity (%): 66% -> 82% | Improved |
| replace FDC with NSE (noised state space exploration) | FDC · GEOM-Drugs unconstrained | diversity and validity: 508 -> 519; 66% -> 74% | Improved |
| add projection step (eta > 0) to NSE giving L-FE (ablation) | NSE · GEOM-Drugs conformers | diversity and validity | Improved |
| add weak-verifier prior KL term (L-FE) over G-FE (ablation) | G-FE (alpha_t = 0) · GEOM-Drugs conformers | diversity and validity | Mixed |
| L-FE versus FDC compute | FDC · GEOM-Drugs | runtime | No clear effect |
Closest papers by meaning
- Active Flow Expansion for Out-of-Distribution Discovery: from Theory to Molecules (NeurIPS 2026 · similarity 0.90)
- Flow Density Control: Generative Optimization Beyond Entropy-Regularized Fine-Tuning (NeurIPS 2025 · similarity 0.89)
- Navigating Chemical Space with Latent Flows (NeurIPS 2024 · similarity 0.88)
- AlphaFold Meets Flow Matching for Generating Protein Ensembles (ICML 2024 · similarity 0.87)
- Unlocking Guidance for Discrete State-Space Diffusion and Flow Models (ICLR 2025 · similarity 0.86)
- Efficient Molecular Conformer Generation with SO(3)-Averaged Flow Matching and Reflow (ICML 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.
