Research map / Learning theory and algorithms / Sample complexity
A Unified View of Label Shift Estimation
Saurabh Garg, Yifan Wu, Sivaraman Balakrishnan, Zachary Lipton
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 fine-grained BCTS calibration instead of confusion-matrix calibration in MLLS (ablation) | MLLS · MNIST, CIFAR10, GMM label shift | MSE of weights: 2-10x lower MSE | Improved |
| replace BBSE moment-matching objective with MLLS likelihood objective, same confusion-matrix calibration (MLLS-CM vs BBSE) (ablation) | BBSE · GMM, MNIST, CIFAR10 | MSE of weights | No clear effect |
| increase number of calibration bins (finer calibration) (ablation) | MLLS · Synthetic GMM | MSE / min Hessian eigenvalue | Improved |
| MLLS vs RLLS and BBSE under severe shift | MLLS with BCTS · MNIST, CIFAR10 | MSE of weights | Improved |
Closest papers by meaning
- LaSCal: Label-Shift Calibration without target labels (NeurIPS 2024 · similarity 0.87)
- Graph–Smoothed Bayesian Black-Box Shift Estimator and Its Information Geometry (NeurIPS 2025 · similarity 0.87)
- Learning Label Shift Correction for Test-Agnostic Long-Tailed Recognition (ICML 2024 · similarity 0.86)
- Open Set Label Shift with Test Time Out-of-Distribution Reference (CVPR 2025 · similarity 0.86)
- Heterogeneous Label Shift: Theory and Algorithm (ICML 2025 · similarity 0.85)
- Hypothesis Testing for Class-Conditional Noise Using Local Maximum Likelihood (AAAI 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.
