Research map / Learning theory and algorithms / Prediction and decision losses
A Forward Approach for Sufficient Dimension Reduction in Binary Classification
Jongkyeong Kang, Seung Jun Shin
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 wOPG-SVM instead of the inverse-method PWSVM | SDR for binary classification · Simulation Model (I), n=1000, p=10 | subspace distance d(B_hat,B): 0.56 -> 0.10 | Improved |
| use wOPG-SVM instead of PWSVM when boundary is symmetric | SDR for binary classification · Simulation Model (V), n=500, p=10 | subspace distance d(B_hat,B): 1.59 -> 0.14 | Improved |
| use wOPG-LR instead of PRE under symmetric boundary | SDR for binary classification · Simulation Model (V), n=1000, p=20 | subspace distance d(B_hat,B): 1.79 -> 0.18 | Improved |
| add the refinement of the full wOPG over the naive gradient estimator (SVM loss) (ablation) | wOPG-SVM · Simulation Model (I), n=1000, p=10 | subspace distance d(B_hat,B): 0.48 -> 0.10 | Improved |
| add the refinement of the full wOPG over the naive gradient estimator (LR loss) (ablation) | wOPG-LR · Simulation Model (I), n=500, p=20 | subspace distance d(B_hat,B): 1.32 -> 0.99 | Improved |
| use hinge (SVM) loss instead of logistic loss in wOPG | wOPG · Simulation Model (I), n=500, p=10 | subspace distance d(B_hat,B): 0.51 -> 0.34 | Improved |
| use hinge (SVM) loss instead of logistic loss in wOPG | wOPG · Simulation Model (II), n=1000, p=20 | subspace distance d(B_hat,B): 0.90 -> 0.67 | Improved |
| use hinge (SVM) loss instead of logistic loss in wOPG | wOPG · Simulation Model (V), n=500, p=10 | subspace distance d(B_hat,B): 0.14 -> 0.14 | No clear effect |
| increase sample size from 500 to 1000 (p=10) | wOPG-SVM · Simulation Model (II) | subspace distance d(B_hat,B): 0.45 -> 0.11 | Improved |
| increase dimension from p=10 to p=20 (n=500) | wOPG-SVM · Simulation Model (II) | subspace distance d(B_hat,B): 0.45 -> 1.16 | Hurt |
| use second-moment methods SAVE instead of PWSVM when boundary is symmetric | SAVE · Simulation Model (V), n=500, p=10 | subspace distance d(B_hat,B): 1.59 -> 0.27 | Improved |
Closest papers by meaning
- Neural Networks Perform Sufficient Dimension Reduction (AAAI 2025 · similarity 0.86)
- Large Margin Discriminant Dimensionality Reduction in Prediction Space (NeurIPS 2016 · similarity 0.86)
- Unified Binary and Multiclass Margin-Based Classification (JMLR 2024 · similarity 0.86)
- The Maximum Separation Subspace in Sufficient Dimension Reduction with Categorical Response (JMLR 2020 · similarity 0.86)
- Deep Principal Support Vector Machines for Nonlinear Sufficient Dimension Reduction (ICML 2025 · similarity 0.85)
- The Implicit Bias of Gradient Descent on Separable Multiclass Data (NeurIPS 2024 · similarity 0.84)
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.
