Research map / Label-efficient and robust learning / Recognition and long-tail
LOTTERY: Learning from Reference-Only Samples in Two-Sample Testing under Size Asymmetry
Xunye Tian, Zhijian Zhou, Liuhua Peng, Feng Liu
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 |
|---|---|---|---|
| add uncertainty-guided selection/weighting of RDRs (ablation) | LOTTERY · BLOB (N=4000, M=20) | test power: 0.411 -> 0.524 | Improved |
| add uncertainty-guided selection/weighting of RDRs (ablation) | LOTTERY · CIFAR10-RES18 (N=100, M=4) | test power: 0.916 -> 0.966 | Improved |
| add uncertainty-guided selection with large query size (ablation) | LOTTERY · BLOB (N=4000, M=300) | test power: 0.995 -> 1.000 | No clear effect |
| aggregate multiple RDR families instead of a single RDR (ablation) | LOTTERY · Synthetic alternatives (N=4000, M=50) | test power: 0.320 -> 0.442 (mean shift, vs Mahalanobis) | Improved |
| learn representations from reference only instead of splitting query data | LOTTERY vs split-based tests · CIFAR10-RES18 (N=100, M=2) | test power: 0.810 vs 0.105 (RL-TST) | Improved |
| increase reference size N with few queries | LOTTERY · BLOB (M=50) | test power: 0.082 -> 0.725 | Improved |
| very small reference set (N=100) | LOTTERY · BLOB (M=50, N=100) | test power: 0.082 vs 0.122 (MMD-FUSE) | Hurt |
| increase reference size for baselines | MMD-FUSE · BLOB (M=50) | test power: 0.122 -> 0.145 | No clear effect |
| weight RDRs by sensitivity divided by uncertainty rather than inverse uncertainty alone | LOTTERY · design argument | test power | No clear effect |
Closest papers by meaning
- DUAL: Learning Diverse Kernels for Aggregated Two-sample and Independence Testing (NeurIPS 2025 · similarity 0.88)
- Revealing Distribution Discrepancy by Sampling Transfer in Unlabeled Data (NeurIPS 2024 · similarity 0.86)
- Kernel Selection is Model Selection: A Unified Complexity-Penalised Approach for MMD Two-Sample Tests (NeurIPS 2026 · similarity 0.85)
- Optimal Algorithms for Augmented Testing of Discrete Distributions (NeurIPS 2024 · similarity 0.85)
- Protected Test-Time Adaptation via Online Entropy Matching: A Betting Approach (NeurIPS 2024 · similarity 0.85)
- Harnessing Hierarchical Label Distribution Variations in Test Agnostic Long-tail Recognition (ICML 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.
