Research map / Learning theory and algorithms / Sample complexity
The Maximum Separation Subspace in Sufficient Dimension Reduction with Categorical Response
Xin Zhang, Qing Mai, Hui Zou
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 MASES instead of SIR | SDR estimators · Inverse models MDA-1 | angle (degrees): 64.7 -> 16.3 | Improved |
| use MASES instead of SIR | SDR estimators · Inverse model MDA-2 | angle (degrees): 53.3 -> 10.4 | Improved |
| use MASES instead of SIR on correlated-predictor LDA models | SDR estimators · LDA-1 Cor. | angle (degrees): 52.7 -> 27.0 | Improved |
| use MASES instead of PSVM/PRE when classes are almost completely separated and Sigma = I | SDR estimators · LDA-2 Ind. | angle (degrees): 23.3 -> 34.4 | Hurt |
| use MASES instead of SIR on forward single-index models | SDR estimators · SIM-1 | angle (degrees): 80.6 -> 37.8 | Improved |
| use MASES instead of logistic regression with independent predictors | SDR estimators · LR-Ind. | angle (degrees): 20.1 -> 22.2 | Hurt |
| use MASES instead of SIR with correlated predictors in logistic model | SDR estimators · LR-Cor. | angle (degrees): 50.1 -> 29.2 | Improved |
| use MASES instead of the best competitor on two-dimensional models | SDR estimators · MIM-3 | Frobenius estimation error: 0.79 -> 0.29 | Improved |
| use MASES instead of SIR on two-dimensional LDA with correlated predictors | SDR estimators · LDA Cor. (2D) | Frobenius estimation error: 1.62 -> 0.41 | Improved |
| apply MASES dimension selection procedure | MASES · Two-dimensional models, 100 replicates | correct dimension selection: 65 of 100 on MIM-1 vs 100 on LDA Ind. | Mixed |
| use MASES instead of SAVE/DR subspaces for class separation | SDR estimators · Birds, planes and cars data | overall Hellinger distance: 0.133 -> 0.982 | Improved |
Closest papers by meaning
- Neural Networks Perform Sufficient Dimension Reduction (AAAI 2025 · similarity 0.89)
- Kernel Sufficient Dimension Reduction and Variable Selection for Compositional Data via Amalgamation (ICML 2023 · similarity 0.89)
- Sufficient dimension reduction for classification using principal optimal transport direction (NeurIPS 2020 · similarity 0.88)
- Enhancing Sufficient Dimension Reduction via Hellinger Correlation (ICML 2024 · similarity 0.88)
- Factor-analytic inverse regression for high-dimension, small-sample dimensionality reduction (ICML 2021 · similarity 0.87)
- Heterogeneous Sufficient Dimension Reduction and Subspace Clustering (ICML 2025 · 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.
