Research map / Label-efficient and robust learning / Recognition and long-tail
Confidence Sets with Expected Sizes for Multiclass Classification
Christophe Denis, Mohamed Hebiri
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 |
|---|---|---|---|
| replace single-label prediction by random forest beta-set with beta=2 | random forest · Plant (K=100) | risk: 0.40 -> 0.18 | Improved |
| aggregate library by superlearning instead of best single beta-set | confidence sets · Plant (K=100), beta=2 | risk R: 0.20 -> 0.17 | Improved |
| use CV aggregation instead of softmax regression beta-set | confidence sets · Plant (K=100), beta=2 | risk R: 0.77 -> 0.17 | Improved |
| increase beta from 2 to 10 | CV confidence set · Plant (K=100) | risk R: 0.17 -> 0.02 | Improved |
| increase beta from 2 to 10 | random forest beta-set · Plant (K=100) | risk R: 0.18 -> 0.02 | Improved |
| use CV aggregation instead of best library component | superlearning CV · Forest (K=4), beta=2 | risk R: 0.02 -> 0.02 | No clear effect |
| use CV aggregation instead of softmax regression beta-set | confidence sets · Forest (K=4), beta=2 | risk R: 0.06 -> 0.02 | Improved |
| calibrate threshold with G estimated from unlabeled data | beta-set · Forest and Plant | information I | Improved |
Closest papers by meaning
- Knowing what You Know: valid and validated confidence sets in multiclass and multilabel prediction (JMLR 2021 · similarity 0.88)
- Demystifying the Optimal Performance of Multi-Class Classification (NeurIPS 2023 · similarity 0.87)
- Theoretical Insights Into Multiclass Classification: A High-dimensional Asymptotic View (NeurIPS 2020 · similarity 0.86)
- On Truthing Issues in Supervised Classification (JMLR 2024 · similarity 0.86)
- Enhanced Sample Selection with Confidence Tracking: Identifying Correctly Labeled Yet Hard-to-Learn Samples in Noisy Data (AAAI 2025 · similarity 0.85)
- Don’t Label Twice: Quantity Beats Quality when Comparing Binary Classifiers on a Budget (ICML 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.
