Research map / Learning theory and algorithms / Optimal transport
Oversampling for Imbalanced Data via Optimal Transport
Yuguang Yan, Mingkui Tan, Yanwu Xu, Jiezhang Cao, Michael Ng, Huaqing Min, Qingyao Wu
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 |
|---|---|---|---|
| OTOS (optimal transport oversampling) instead of SMOTE | linear SVM · svmguide4 | G-mean: 0.6719 -> 0.8037 | Improved |
| oversample with OTOS instead of no oversampling | linear SVM · diabetes | G-mean: 0.6953 -> 0.7495 | Improved |
| OTOS instead of SMOTE | linear SVM · iSee-DR | G-mean: 0.4754 -> 0.5014 | Improved |
| shift minority distribution by hinge loss (OTOS-beta vs OTOS-alpha) (ablation) | OTOS-alpha · diabetes | G-mean: 0.7443 -> 0.7466 | Improved |
| shift minority distribution by hinge loss (OTOS-beta vs OTOS-alpha) (ablation) | OTOS-alpha · svmguide4 | G-mean: 0.7758 -> 0.7807 | Improved |
| add regularization on transported samples vs majority samples (OTOS vs OTOS-beta) (ablation) | OTOS-beta · german | G-mean: 0.7204 -> 0.7255 | Improved |
| add regularization on transported samples vs majority samples (OTOS vs OTOS-beta) (ablation) | OTOS-beta · svmguide4 | G-mean: 0.7807 -> 0.8037 | Improved |
| add regularization on transported samples vs majority samples (OTOS vs OTOS-beta) (ablation) | OTOS-beta · iSee-DR | G-mean: 0.5034 -> 0.5014 | Mixed |
| add loss shift and regularization (OTOS vs OTOS-alpha) (ablation) | OTOS-alpha · iSee-DR | G-mean: 0.4891 -> 0.5014 | Improved |
| OTOS vs ADASYN on sensitivity | linear SVM · benchmark data sets | Sensitivity win/tie/loss: 5/3/2 | Mixed |
Closest papers by meaning
- Concentration and excess risk bounds for imbalanced classification with synthetic oversampling (NeurIPS 2025 · similarity 0.86)
- Restoring balance: principled under/oversampling of data for optimal classification (ICML 2024 · similarity 0.86)
- Balancing the Scales: A Theoretical and Algorithmic Framework for Learning from Imbalanced Data (ICML 2025 · similarity 0.83)
- Learning under Localized Minority Imbalance (NeurIPS 2026 · similarity 0.83)
- Exploiting Geometry for Treatment Effect Estimation via Optimal Transport (AAAI 2024 · similarity 0.82)
- A Unified Loss for Handling Inter-Class and Intra-Class Imbalance in Medical Image Segmentation (AAAI 2025 · similarity 0.82)
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.
