Research map / Probabilistic and scientific modeling / Variational and generative sampling
Flow Stochastic Segmentation Networks
Fabio De Sousa Ribeiro, Omar Todd, Charles Jones, Avinash Kori, Raghav Mehta, Ben Glocker
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 continuous-time flow instead of the SSN++ baseline | Flow-SSN · LIDC-IDRI | D2_GED (M samples): 0.212 -> 0.182 | Improved |
| Use continuous-time flow instead of SSN++ | Flow-SSN · LIDC-IDRI | Dice: 0.471 -> 0.610 | Improved |
| Use discrete single-step IAF Flow-SSN instead of SSN++ | Flow-SSN · LIDC-IDRI | D2_GED (M samples): 0.212 -> 0.212 | No clear effect |
| Learn the base distribution (prior) instead of holding it fixed (ablation) | Continuous Flow-SSN · LIDC-IDRI | inference time: 5x reduction in inference time | Improved |
| Learn the prior so the flow network is much smaller (ablation) | Continuous Flow-SSN · LIDC-IDRI | parameters in flow network: 150K vs 14M params | Improved |
| Reduce Euler ODE steps from 250 to 10 (ablation) | Continuous Flow-SSN · LIDC-IDRI | D2_GED: T=10 competitive | No clear effect |
| Increase Euler steps from 50 to 250 (ablation) | Continuous Flow-SSN · LIDC-IDRI | D2_GED (16 samples): 0.210 -> 0.207 | No clear effect |
| Use Dopri5 adaptive solver instead of Euler T=250 (ablation) | Continuous Flow-SSN · LIDC-IDRI | D2_GED (16 samples): 0.209 -> 0.207 | No clear effect |
| Use Flow-SSN instead of the SSN with low-rank covariance | Flow-SSN · MarkovShapes | bits per dimension: outperforms all SSN variants, including full-rank | Improved |
| Use low rank 2 SSN versus Flow-SSN (ablation) | SSN · MarkovShapes | sample quality: rank 2 SSN hallucinates nonexistent shapes | Hurt |
| Increase MC samples from 16 to 512 (ablation) | Discrete Flow-SSN · REFUGE-MultiRater optic cup | D2_GED: 0.116 -> 0.081 | Improved |
| Increase MC samples from 16 to 512 (ablation) | Discrete Flow-SSN · REFUGE-MultiRater optic cup | HM-IoU: 0.766 -> 0.881 | Improved |
Closest papers by meaning
- MedFlowSeg: Flow Matching for Medical Image Segmentation with Frequency-Aware Attention (NeurIPS 2026 · similarity 0.89)
- Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation (NeurIPS 2025 · similarity 0.86)
- Unifying Bayesian Flow Networks and Diffusion Models through Stochastic Differential Equations (ICML 2024 · similarity 0.86)
- Rethinking Vector Field Learning for Generative Segmentation (NeurIPS 2026 · similarity 0.85)
- Asymmetric Flow Models (NeurIPS 2026 · similarity 0.85)
- STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation (NeurIPS 2026 · similarity 0.85)
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.
