Research map / Label-efficient and robust learning / Recognition and long-tail
Predicting Classification Accuracy When Adding New Unobserved Classes
Yuli Slavutsky, Yuval Benjamini
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 CleaneX instead of regression-based prediction | Accuracy extrapolation · CIFAR-100 | RMSE: 18% | Improved |
| Use CleaneX instead of regression-based prediction | Accuracy extrapolation · LFW | RMSE: 32% | Improved |
| Use CleaneX instead of regression-based prediction | Accuracy extrapolation · fMRI | RMSE: 22% | Improved |
| Use CleaneX instead of KDE | Accuracy extrapolation · CIFAR-100 | RMSE: 7% | Improved |
| Use CleaneX instead of KDE | Accuracy extrapolation · LFW | RMSE: 38% | Hurt |
| Use CleaneX instead of KDE | Accuracy extrapolation · fMRI | RMSE: 73% | Improved |
| Use CleaneX instead of KDE | Accuracy extrapolation · Simulations, d=5 | RMSE | Improved |
| Use CleaneX instead of regression-based prediction | Accuracy extrapolation · Simulations, d=5, k1=100 | RMSE | Improved |
| Use CleaneX instead of regression-based prediction | Accuracy extrapolation · Simulations, d=5, k1=500 | RMSE | No clear effect |
| Use CleaneX instead of both competing methods | Accuracy extrapolation · Simulations, d=3 | RMSE | Improved |
| Use CleaneX instead of regression-based prediction | Accuracy extrapolation · Simulations, d=10 | RMSE | No clear effect |
| Calibrate score-based estimates using observed accuracies | CleaneX · CIFAR-100 | RMSE: 7% | Improved |
Closest papers by meaning
- Extrapolating Expected Accuracies for Large Multi-Class Problems (JMLR 2018 · similarity 0.89)
- Estimating Accuracy from Unlabeled Data: A Bayesian Approach (ICML 2016 · similarity 0.86)
- Theoretical Insights Into Multiclass Classification: A High-dimensional Asymptotic View (NeurIPS 2020 · similarity 0.85)
- Demystifying the Optimal Performance of Multi-Class Classification (NeurIPS 2023 · similarity 0.85)
- A Data-Driven Measure of Relative Uncertainty for Misclassification Detection (ICLR 2024 · similarity 0.85)
- The VOROS: Lifting ROC Curves to 3D to Summarize Unbalanced Classifier Performance (AAAI 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.
