Research map / Learning theory and algorithms / Prediction and decision losses
Variable-Based Calibration for Machine Learning Classifiers
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 |
|---|---|---|---|
| Calibrate separately within Age-based tree leaves instead of using beta calibration globally | Adult Income MLP · Adult Income (Age) | VECE: 9.59% -> 2.11% | Helped |
| Calibrate separately within Age-based tree leaves instead of using beta calibration globally | Adult Income MLP · Adult Income (Age) | ECE: 1.65% -> 1.64% | Helped |
| Calibrate separately within review-length tree leaves instead of using beta calibration globally | Yelp BERT · Yelp (review length) | VECE: 0.37% -> 0.23% | Helped |
| Calibrate separately within review-length tree leaves instead of using beta calibration globally | Yelp BERT · Yelp (review length) | ECE: 1.73% -> 1.70% | Helped |
| Calibrate separately within Age-based tree leaves instead of using Platt scaling globally | Bank Marketing MLP · Bank Marketing (Age) | VECE: 2.83% -> 0.52% | Helped |
| Calibrate separately within Age-based tree leaves instead of using Platt scaling globally | Bank Marketing MLP · Bank Marketing (Age) | ECE: 2.38% -> 2.10% | Helped |
| Calibrate separately within reaction-time tree leaves instead of using Dirichlet calibration globally | CIFAR-10H DenseNet · CIFAR-10H (median reaction time) | VECE: 1.12% -> 0.86% | Helped |
| Calibrate separately within reaction-time tree leaves instead of using Dirichlet calibration globally | CIFAR-10H DenseNet · CIFAR-10H (median reaction time) | ECE: 0.80% -> 1.18% | Hurt |
| Apply Platt scaling | Yelp BERT · Yelp (review length) | VECE: 1.93% -> 0.64% | Hurt |
| Use augmented-beta variable-based calibration instead of tree-based variable-based calibration | Adult Income MLP · Adult Income (Age) | VECE: 2.11% -> 1.87% | Helped |
| Use augmented-beta variable-based calibration instead of tree-based variable-based calibration | Yelp BERT · Yelp (review length) | VECE: 0.23% -> 0.37% | Hurt |
| Use augmented-beta variable-based calibration instead of tree-based variable-based calibration | Bank Marketing MLP · Bank Marketing (Age) | VECE: 0.52% -> 1.13% | Hurt |
Closest papers by meaning
- Reassessing How to Compare and Improve the Calibration of Machine Learning Models (ICLR 2025 · similarity 0.90)
- Divide et Calibra: Multiclass Local Calibration via Vector Quantization (NeurIPS 2026 · similarity 0.86)
- Testing Calibration in Nearly-Linear Time (NeurIPS 2024 · similarity 0.86)
- How Flawed Is ECE? An Analysis via Logit Smoothing (ICML 2024 · similarity 0.86)
- LaSCal: Label-Shift Calibration without target labels (NeurIPS 2024 · similarity 0.86)
- Combining Priors with Experience: Confidence Calibration Based on Binomial Process Modeling (AAAI 2025 · 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.