Research map / Label-efficient and robust learning / Recognition and long-tail
Classification with Rejection Based on Cost-sensitive Classification
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 CS-hinge | Cost-sensitive approach · Clean-labeled binary classification | test empirical zero-one-c risk | Helped |
| use CS-sigmoid | Cost-sensitive approach · Noisy-labeled binary classification | test empirical zero-one-c risk | Helped |
| use CS-sigmoid | Cost-sensitive approach · Positive-unlabeled binary classification | test empirical zero-one-c risk | Helped |
| use CS-sigmoid | Cost-sensitive approach · Noisy-labeled multiclass classification | test empirical zero-one-c risk | Helped |
| apply uniform label noise | Classification with rejection methods · Noisy-labeled classification | test empirical zero-one-c risk | Hurt |
| use SCE with temperature scaling | SCE · Clean-labeled multiclass classification | test empirical zero-one-c risk | No clear effect |
| use SCE, even with temperature scaling | SCE · Clean-labeled multiclass classification | rejection of ambiguous data | Hurt |
| use DEFER | Classification with rejection · Binary classification | test empirical zero-one-c risk | Mixed |
| use ANGLE | Classification with rejection · Gas-Drift and HAR | test empirical zero-one-c risk | Helped |
| use ANGLE | Classification with rejection · MNIST, Fashion-MNIST, and KMNIST | test empirical zero-one-c risk | Hurt |
| use SCE or ANGLE in PU-classification | SCE and ANGLE · Positive-unlabeled classification | test empirical zero-one-c risk | Hurt |
Closest papers by meaning
- On the Calibration of Multiclass Classification with Rejection (NeurIPS 2019 · similarity 0.88)
- On discriminative learning of prediction uncertainty (ICML 2019 · similarity 0.87)
- Generalizing Consistent Multi-Class Classification with Rejection to be Compatible with Arbitrary Losses (NeurIPS 2022 · similarity 0.87)
- Regression with Cost-based Rejection (NeurIPS 2023 · similarity 0.86)
- Active Learning for Cost-Sensitive Classification (ICML 2017 · similarity 0.85)
- Learning with Selectively Labeled Data from Multiple Decision-makers (ICML 2025 · similarity 0.83)
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.