Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Modeling Multimodal Aleatoric Uncertainty in Segmentation with Mixture of Stochastic Experts
Zhitong Gao, Yucong Chen, Chuyu Zhang, Xuming 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 weighted-sample compact representation instead of standard sampling | MoSE · LIDC-IDRI, all labels | GED (16): 0.218 $\pm$ 0.003 -> 0.195 $\pm$ 0.005 | Improved |
| Use weighted-sample compact representation instead of standard sampling | MoSE · LIDC-IDRI, all labels | M-IoU (16): 0.624 $\pm$ 0.004 -> 0.635 $\pm$ 0.003 | Improved |
| Use weighted-sample compact representation instead of standard sampling | MoSE · Multimodal Cityscapes, additional input, 35 samples | M-IoU: 0.690 $\pm$ 0.002 -> 0.761 $\pm$ 0.002 | Improved |
| Replace OT-based loss with pairwise IoU loss (ablation) | MoSE · LIDC-IDRI, all labels | GED: 0.218 $\pm$ 0.003 -> 0.533 $\pm$ 0.001 | Hurt |
| Replace OT-based loss with pairwise IoU loss (ablation) | MoSE · LIDC-IDRI, all labels | M-IoU: 0.624 $\pm$ 0.004 -> 0.533 $\pm$ 0.001 | Hurt |
| Replace OT-based loss with pairwise IoU loss (ablation) | MoSE · LIDC-IDRI, all labels | ECE: 0.064 $\pm$ 0.015 -> 0.277 $\pm$ 0.017 | Hurt |
| Use uniform instead of learnable expert weights (ablation) | MoSE · LIDC-IDRI, all labels | GED: 0.218 $\pm$ 0.003 -> 0.282 $\pm$ 0.002 | Hurt |
| Use uniform instead of learnable expert weights (ablation) | MoSE · LIDC-IDRI, all labels | M-IoU: 0.624 $\pm$ 0.004 -> 0.545 $\pm$ 0.007 | Hurt |
| Use uniform instead of learnable expert weights (ablation) | MoSE · LIDC-IDRI, all labels | ECE: 0.064 $\pm$ 0.015 -> 0.215 $\pm$ 0.006 | Hurt |
| Replace stochastic experts with deterministic experts (ablation) | MoSE · LIDC-IDRI, all labels | GED: 0.218 $\pm$ 0.003 -> 0.246 $\pm$ 0.006 | Hurt |
| Replace stochastic experts with deterministic experts (ablation) | MoSE · LIDC-IDRI, all labels | M-IoU: 0.624 $\pm$ 0.004 -> 0.591 $\pm$ 0.001 | Hurt |
| Replace stochastic experts with deterministic experts (ablation) | MoSE · LIDC-IDRI, all labels | ECE: 0.064 $\pm$ 0.015 -> 0.142 $\pm$ 0.003 | Hurt |
Closest papers by meaning
- Structuring Uncertainty for Fine-Grained Sampling in Stochastic Segmentation Networks (NeurIPS 2022 · similarity 0.89)
- ValUES: A Framework for Systematic Validation of Uncertainty Estimation in Semantic Segmentation (ICLR 2024 · similarity 0.88)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.87)
- UncertainSAM: Fast and Efficient Uncertainty Quantification of the Segment Anything Model (ICML 2025 · similarity 0.87)
- Predictive Uncertainty Quantification for Bird's Eye View Segmentation: A Benchmark and Novel Loss Function (ICLR 2025 · similarity 0.86)
- UNICL-SAM: Uncertainty-Driven In-Context Segmentation with Part Prototype Discovery (CVPR 2025 · 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.
