Research map / Label-efficient and robust learning / Recognition and long-tail
An Unbiased Risk Estimator for Learning with Augmented Classes
Yu-Jie Zhang, Peng Zhao, Lanjihong Ma, Zhi-Hua Zhou
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 unbiased risk estimator with unlabeled data (EULAC) instead of best geometric baseline | RKHS classifiers · SenseVeh | Macro-F1: 54.12 -> 77.33 | Improved |
| use EULAC instead of best baseline | RKHS classifiers · usps | Macro-F1: 79.77 -> 86.52 | Improved |
| use EULAC instead of EVM | RKHS classifiers · pendigits | Macro-F1: 89.94 -> 88.41 | Hurt |
| use DNN-based EULAC instead of OSRCI | DNN open-set · Cifar-10 | AUC of augmented class: 69.9 -> 85.2 | Improved |
| use DNN-based EULAC instead of OSRCI | DNN open-set · mnist | AUC of augmented class: 98.8 -> 98.6 | No clear effect |
| increase unlabeled data size from 250 to 1500 (ablation) | EULAC · mnist, landset, usps | Macro-F1 | Improved |
| increase unlabeled data size (ablation) | LACU-SVM · mnist, landset, usps | Macro-F1 | No clear effect |
| misspecify the mixture proportion theta (ablation) | EULAC · mnist | performance | Hurt |
| account for known-class prior shift (EULAC-shift vs base) (ablation) | EULAC · mnist with prior shift | accuracy | Improved |
Closest papers by meaning
- A Generalized Unbiased Risk Estimator for Learning with Augmented Classes (AAAI 2023 · similarity 0.95)
- Class Prior Estimation with Biased Positives and Unlabeled Examples (AAAI 2020 · similarity 0.86)
- Unsupervised Risk Estimation Using Only Conditional Independence Structure (NeurIPS 2016 · similarity 0.85)
- Transformation-Invariant Learning and Theoretical Guarantees for OOD Generalization (NeurIPS 2024 · similarity 0.84)
- On the Provable Advantage of Unsupervised Pretraining (ICLR 2024 · similarity 0.84)
- Class Distribution Shifts in Zero-Shot Learning: Learning Robust Representations (NeurIPS 2024 · 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.
