Research map / Label-efficient and robust learning / Recognition and long-tail
CONSIGN: Conformal Segmentation Informed by Spatial Groupings via Decomposition
Bruno Viti, Elias Karabelas, Martin Holler
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 spatially-aware PCA prediction sets instead of pixel-wise RAPS sets | CONSIGN · M&Ms-2, MS-CMR19, LIDC, COCO | Chao estimator (set volume) | Improved |
| use CONSIGN instead of RAPS and SACP on LIDC | CONSIGN · LIDC | Chao estimator | Improved |
| use CONSIGN instead of RAPS and SACP | CONSIGN · COCO-vehicles | sample-based estimated coverage | Improved |
| increase principal components from K=2 to K=5 (ablation) | CONSIGN · COCO | Chao estimator | Hurt |
| use probabilistic U-Net samples instead of dropout samples | CONSIGN · LIDC vs cardiac datasets | Chao estimator | Improved |
| use SACP neighborhood aggregation instead of pixel-wise | RAPS · segmentation datasets | volume, sEC, correlation | No clear effect |
| use CONSIGN instead of pixel-wise sampling | CONSIGN · segmentation datasets | average Pearson correlation | Improved |
| use CONSIGN calibration instead of RAPS | CONSIGN · all datasets | offline calibration time (min): ~0.1-1 -> ~5-15 | Hurt |
Closest papers by meaning
- COMPASS: Robust Feature Conformal Prediction for Medical Segmentation Metrics (ICLR 2026 · similarity 0.89)
- Kandinsky Conformal Prediction: Efficient Calibration of Image Segmentation Algorithms (CVPR 2024 · similarity 0.88)
- Are We Overconfident in Models and Results for Semi-Supervised 3D Medical Image Segmentation? (ICML 2026 · similarity 0.87)
- QUTCC: Quantile Uncertainty Training and Conformal Calibration for Imaging Inverse Problems (NeurIPS 2026 · similarity 0.85)
- DyCON: Dynamic Uncertainty-aware Consistency and Contrastive Learning for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.85)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · 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.
