Research map / Learning theory and algorithms / Prediction and decision losses
Fast Epigraphical Projection-based Incremental Algorithms for Wasserstein Distributionally Robust Support Vector Machine
Jiajin Li, Caihua Chen, Anthony Man-Cho So
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 IPPA instead of ISG (more aggressive shrinking rate rho) | incremental DRSVM solver · synthetic data, sharpness instances | convergence speed | Improved |
| use IPPA instead of ISG on instances without sharpness | incremental DRSVM solver · synthetic non-sharp instances | convergence | Improved |
| vary mini-batch size of ISG (ablation) | M-ISG · synthetic data | convergence speed | Mixed |
| add IPPA refinement after mini-batch ISG (Hybrid) | M-ISG · a1a l1-DRSVM | objective value: 0.651090 -> 0.651090 | Improved |
| add IPPA refinement after mini-batch ISG (Hybrid) | M-ISG · a3a l1-DRSVM | objective value: 0.662962 -> 0.662962 | Improved |
| use Hybrid instead of M-ISG | l1-DRSVM · a1a | wall-clock time: 0.706 -> 1.560 | Hurt |
| use Hybrid instead of IPPA | l1-DRSVM · a1a | wall-clock time: 6.1242 -> 1.560 | Improved |
| use Hybrid instead of YALMIP solver | l1-DRSVM · a9a | wall-clock time: 155.980 -> 3.063 | Improved |
| use IPPA instead of YALMIP solver | l2-DRSVM · ijcnn | wall-clock time: 379.943 -> 20.567 | Improved |
| use IPPA instead of YALMIP solver | l2-DRSVM · a1a | objective value: 0.6339472 vs 0.6338819 | Hurt |
| use Hybrid instead of GS-ADMM | l-inf-DRSVM · a1a | wall-clock time: 5.939 -> 4.789 | Improved |
| use Hybrid instead of GS-ADMM | l-inf-DRSVM · a3a | wall-clock time: 9.638 -> 16.252 | Hurt |
Closest papers by meaning
- A First-Order Algorithmic Framework for Distributionally Robust Logistic Regression (NeurIPS 2019 · similarity 0.86)
- Projection Robust Wasserstein Barycenters (ICML 2021 · similarity 0.85)
- Wasserstein Distributionally Robust Optimization through the Lens of Structural Causal Models and Individual Fairness (NeurIPS 2024 · similarity 0.85)
- Outlier-Robust Wasserstein DRO (NeurIPS 2023 · similarity 0.84)
- Universal generalization guarantees for Wasserstein distributionally robust models (ICLR 2025 · similarity 0.82)
- Wasserstein Gradient Boosting: A Framework for Distribution-Valued Supervised Learning (NeurIPS 2024 · 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.
