Research map / Large language models / Language, safety and interpretability
Optimized Deferral for Imbalanced Settings
Corinna Cortes, Anqi Mao, Mehryar Mohri, Yutao Zhong
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 margin-based MILD loss instead of TDEF | two-stage deferral · SVHN synthetic experts (Setup I, error) | deferral loss: 0.0468 -> 0.0254 | Improved |
| use MILD instead of TDEF | two-stage deferral · CIFAR-10 synthetic experts (Setup I, error) | deferral loss: 0.0520 -> 0.0403 | Improved |
| use MILD instead of TDEF | two-stage deferral · real experts (CIFAR-100 Setup I, error) | deferral loss: 0.4368 -> 0.4265 | Improved |
| use MILD instead of TDEF | two-stage deferral · severe expert imbalance (CIFAR-10) | deferral loss: 0.0810 -> 0.0125 | Improved |
| use MILD instead of TDEF | LLM router · MMLU routing, error only | deferral loss: 0.438 -> 0.425 | Improved |
| use MILD instead of TDEF | LLM router · MMLU routing, error + cost | deferral loss: 0.928 -> 0.813 | Improved |
| use MILD instead of CE, CWCE, LDAM | LLM router · MMLU routing, error + cost | deferral loss: 0.985 / 0.942 / 0.915 -> 0.813 | Improved |
| use theory-derived rho instead of uniform rho (ablation) | MILD · MMLU routing, error + cost | deferral loss: 0.884 -> 0.822 | Improved |
| tune rho by grid search around theory values (ablation) | MILD · MMLU routing, error + cost | deferral loss: 0.822 -> 0.813 | No clear effect |
| use MILD under 20% noise in cost matrix | MILD vs TDEF · MMLU routing, error + cost | deferral loss: TDEF 0.928 -> 1.054; MILD 0.813 -> 0.826 | Improved |
| use MILD under extreme 95%/5% imbalance | MILD vs TDEF · MMLU routing extreme imbalance | deferral loss: 0.950 -> 0.885 | Improved |
| use MILD instead of TDEF | two-stage deferral · Tiny ImageNet synthetic experts (Setup III, error + cost) | deferral loss: 0.8481 -> 0.8167 | Improved |
Closest papers by meaning
- Mastering Multiple-Expert Routing: Realizable $H$-Consistency and Strong Guarantees for Learning to Defer (ICML 2025 · similarity 0.87)
- Regression with Multi-Expert Deferral (ICML 2024 · similarity 0.86)
- Dist Loss: Enhancing Regression in Few-Shot Region through Distribution Distance Constraint (ICLR 2025 · similarity 0.85)
- Density-Ratio Losses for Post-Hoc Learning to Defer (NeurIPS 2026 · similarity 0.85)
- Balancing the Scales: A Theoretical and Algorithmic Framework for Learning from Imbalanced Data (ICML 2025 · similarity 0.84)
- Addressing Imbalanced Domain-Incremental Learning through Dual-Balance Collaborative Experts (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.
