Research map / Probabilistic and scientific modeling / Causal discovery
C2MIL: Synchronizing Semantic and Topological Causalities in Multiple Instance Learning for Robust and Interpretable Survival Analysis
Min Cen, Zhenfeng Zhuang, Yuzhe Zhang, Min Zeng, Baptiste Magnier, Lequan Yu, Hong Zhang, Liansheng Wang
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 |
|---|---|---|---|
| add contrastive loss (ablation) | C2MIL · TCGA-ESCA | C-index: 0.6252 -> 0.6904 | Improved |
| add contrastive loss (ablation) | C2MIL · TCGA-KIRC | C-index: 0.6779 -> 0.7078 | Improved |
| add contrastive loss (ablation) | C2MIL · TCGA-BLCA | C-index: 0.5783 -> 0.6081 | Improved |
| add disentangling loss (ablation) | C2MIL · TCGA-ESCA | C-index: 0.6542 -> 0.6904 | Improved |
| add ratio loss (ablation) | C2MIL · TCGA-KIRC | C-index: 0.6914 -> 0.7078 | Improved |
| use loss weights 0.1:0.1:0.1 instead of 0.5:0.1:0.1 (ablation) | C2MIL · TCGA-KIRC | C-index: 0.6821 -> 0.7078 | Improved |
| adaptive cluster number instead of fixed K=3 (ablation) | C2MIL · TCGA-ESCA | C-index: 0.6684 -> 0.6904 | Improved |
| adaptive cluster number instead of fixed K=2 (ablation) | C2MIL · TCGA-BLCA | C-index: 0.5905 -> 0.6081 | Improved |
| adaptive clusters vs oracle clusters (ablation) | C2MIL · TCGA-KIRC | C-index: 0.7078 vs 0.7131 | No clear effect |
| add semantic disentangling module (ablation) | ABMIL · TCGA-ESCA | C-index: 0.6385 -> 0.6266 | Hurt |
| add semantic disentangling module (ablation) | ABMIL · TCGA-KIRC | C-index: 0.6794 -> 0.6996 | Improved |
| add semantic disentangling module (ablation) | TransMIL · TCGA-ESCA | C-index: 0.5651 -> 0.6368 | Improved |
Closest papers by meaning
- Symb-xMIL: Symbolic Explanations for Multiple Instance Learning in Digital Pathology (NeurIPS 2026 · similarity 0.86)
- Graph Domain Adaptation with Dual-branch Encoder and Two-level Alignment for Whole Slide Image-based Survival Prediction (ICCV 2025 · similarity 0.86)
- Interventional Causal Discovery in a Mixture of DAGs (NeurIPS 2024 · similarity 0.84)
- CaMIL: Causal Multiple Instance Learning for Whole Slide Image Classification (AAAI 2024 · similarity 0.84)
- Causal Discovery with Fewer Conditional Independence Tests (ICML 2024 · similarity 0.84)
- Causal Abstraction Learning based on the Semantic Embedding Principle (ICML 2025 · similarity 0.84)
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.
