Research map / Learning theory and algorithms / Kernels and regression theory
An Asymptotic Study of Discriminant and Vote-Averaging Schemes for Randomly-Projected Linear Discriminants
Lama B. Niyazi, Abla Kammoun, Hayssam Dahrouj, Mohamed-Slim Alouini, Tareq Y. Al-Naffouri
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 discriminant averaging instead of vote averaging | RP-LDA ensemble · Gaussian mixture synthetic | AUC: 0.8340 -> 0.8435 | Improved |
| add projection selection (B1=50, B2=4) to discriminant averaging (ablation) | RP-LDA ensemble · Gaussian mixture synthetic | AUC: 0.8435 -> 0.8059 | Hurt |
| add projection selection to vote averaging (ablation) | RP-LDA ensemble · Gaussian mixture synthetic | AUC: 0.8340 -> 0.7843 | Hurt |
| use discriminant averaging instead of vote averaging | RP-LDA ensemble · gastro WL | AUC: 0.834 -> 0.813 | Hurt |
| use discriminant averaging instead of vote averaging | RP-LDA ensemble · prostate | AUC: 0.958 -> 0.9493 | Hurt |
| add projection selection to discriminant averaging (ablation) | RP-LDA ensemble · prostate full | AUC: 0.787 -> 0.765 | Hurt |
| add projection selection to vote averaging (ablation) | RP-LDA ensemble · prostate full | AUC: 0.770 -> 0.74 | Hurt |
| add projection selection to discriminant averaging (ablation) | RP-LDA ensemble · phoneme aa ao | AUC: 0.8478 -> 0.8409 | Hurt |
| tune M and d with G-estimators instead of cross-validation | finite disc-avg RP-LDA ensemble · Gaussian mixture synthetic | test error: 0.241 -> 0.220 | Improved |
| tune M and d with G-estimators instead of cross-validation (tuning time) | finite disc-avg RP-LDA ensemble · Gaussian mixture synthetic | tuning time (s): 498.517 -> 8.800 | Improved |
| add the closed-form heuristic for M on top of G-estimator tuning | finite disc-avg RP-LDA ensemble · Gaussian mixture synthetic | test error: 0.220 -> 0.218 | Improved |
| tune with G-estimators instead of Durrant's rule-of-thumb | finite disc-avg RP-LDA ensemble · Gaussian mixture synthetic | test error: 0.245 -> 0.220 | Improved |
Closest papers by meaning
- Random-projection ensemble dimension reduction (ICLR 2026 · similarity 0.86)
- High-dimensional Linear Discriminant Analysis Classifier for Spiked Covariance Model (JMLR 2020 · similarity 0.85)
- How many classifiers do we need? (NeurIPS 2024 · similarity 0.83)
- On Rate-Optimal Partitioning Classification from Observable and from Privatised Data (NeurIPS 2026 · similarity 0.82)
- No Free Lunch from Random Feature Ensembles: Scaling Laws and Near-Optimality Conditions (ICML 2025 · similarity 0.82)
- Theoretical Limitations of Ensembles in the Age of Overparameterization (ICML 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.
