Research map / Label-efficient and robust learning / Recognition and long-tail
Revisiting Classifier Two-Sample Tests
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 C2ST-NN instead of the listed two-sample test baselines | Two-sample testing · NIPS article categories | power | Helped |
| Use C2ST-NN instead of SCF-full or MMD-quad | Two-sample testing · Facial expressions + vs. - | power | Hurt |
| Use C2ST-KNN instead of the strongest listed causal-discovery baseline | Tübingen cause-effect discovery · Tübingen cause-effect pairs | accuracy | Helped |
| Increase classifier flexibility, such as increasing hidden neurons or decreasing nearest neighbors (ablation) | C2ST-NN and C2ST-KNN · Synthetic two-sample tests | performance | No clear effect |
| Evaluate real and generated images on pixels rather than pretrained ResNet features | GAN sample-quality evaluation · LSUN and LFW | test accuracy | Hurt |
| Use a fresh discriminator trained and tested on held-out data rather than the GAN's training discriminator | GAN sample-quality evaluation · GAN evaluation | sample-quality assessment | Helped |
Closest papers by meaning
- Conformal C2ST: Turning weak classifiers into strong two-sample tests (ICML 2026 · similarity 0.91)
- Meta Two-Sample Testing: Learning Kernels for Testing with Limited Data (NeurIPS 2021 · similarity 0.87)
- Don’t Label Twice: Quantity Beats Quality when Comparing Binary Classifiers on a Budget (ICML 2024 · similarity 0.85)
- Synthetic Tabular Data Generation for Imbalanced Classification: The Surprising Effectiveness of an Overlap Class (AAAI 2025 · similarity 0.83)
- Leveraging Uncertainty Estimates To Improve Classifier Performance (ICLR 2024 · similarity 0.83)
- Fusing Conditional Submodular GAN and Programmatic Weak Supervision (AAAI 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.