Research map / Learning theory and algorithms / Prediction and decision losses
On the Calibration of Multiclass Classification with Rejection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the confidence-based CE loss and derive rejection from its inverse-link class probabilities | Multiclass classification with rejection · Synthetic datasets | test 0-1-c risk | Helped |
| Use the confidence-based CE loss instead of OVA+log | Multiclass classification with rejection · Benchmark datasets | test 0-1-c risk | Helped |
| Use the classifier-rejector approach instead of confidence-based methods | Multiclass classification with rejection · Benchmark datasets | test 0-1-c risk | Mixed |
| Use OVA+log rather than CE | Multiclass classification with rejection · Benchmark datasets | test 0-1-c risk | Hurt |
| Use the classifier-rejector approach with MPC+log | Multiclass classification with rejection · Synthetic datasets | rejection calibration | Hurt |
| Use classifier-rejector methods on difficult-to-learn datasets | Multiclass classification with rejection · yeast | rejection behavior | Hurt |
Closest papers by meaning
- Generalizing Consistent Multi-Class Classification with Rejection to be Compatible with Arbitrary Losses (NeurIPS 2022 · similarity 0.90)
- Classification with Rejection Based on Cost-sensitive Classification (ICML 2021 · similarity 0.88)
- Calibration tests in multi-class classification: A unifying framework (NeurIPS 2019 · similarity 0.88)
- Demystifying the Optimal Performance of Multi-Class Classification (NeurIPS 2023 · similarity 0.87)
- Toward Minimal-dimensional Convex Calibrated Surrogate Losses for Classification with Rejection (NeurIPS 2026 · similarity 0.87)
- Confidence Calibration of Classifiers with Many Classes (NeurIPS 2024 · similarity 0.86)
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.