Research map / Learning theory and algorithms / Kernels and regression theory
Diversity-Aware Recursive Feature Multiple Kernel Learning
Nan Cao, Xu Zhao, Teng Zhang
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 |
|---|---|---|---|
| add RFM kernels (learned feature weighting) to the candidate pool instead of RBF and polynomial only (ablation) | DARFMMKL · sonar | test accuracy (%): 82.29 -> 85.90 | Improved |
| add RFM kernels (ablation) | DARFMMKL · ionosphere | test accuracy (%): 91.51 -> 95.07 | Improved |
| add RFM kernels (ablation) | DARFMMKL · german.numer | test accuracy (%): 73.07 -> 76.58 | Improved |
| add traditional kernels on top of RFM kernels (Combined vs RFM-Only) (ablation) | DARFMMKL · splice | test accuracy (%): 85.65 -> 87.32 | Improved |
| add diversity term to CKA quality-only kernel selection (ablation) | DARFMMKL · ionosphere | test accuracy (%): 91.24 -> 93.16 | Improved |
| add diversity term to CKA quality-only kernel selection (ablation) | DARFMMKL · breast-cancer, w7a | test accuracy (%): 96.46 -> 96.46; 97.84 -> 97.84 | No clear effect |
| add diversity term to CKA quality-only kernel selection (ablation) | DARFMMKL · sonar, heart, ionosphere, a8a | test accuracy (%): avg. +0.73 on four datasets where active | Improved |
| increase number of RFM iterations T (ablation) | DARFMMKL · sonar, heart, breast-cancer, ionosphere | test accuracy (%) | No clear effect |
| stratified or K-means vs uniform Nystrom landmark sampling (ablation) | DARFMMKL · sonar, heart, breast-cancer, ionosphere | test accuracy (%): sonar 74.76 -> 79.05 (stratified); heart 81.11 -> 80.21 | Mixed |
| very large lambda (over-weight CKA quality vs diversity) (ablation) | DARFMMKL · 7 datasets | test accuracy | Hurt |
| increase number of selected kernels m beyond optimum (ablation) | DARFMMKL · 7 datasets | test accuracy | Hurt |
| swap downstream learner (SVM, LS-SVM, ODM, ELM) (ablation) | DARFMMKL selected kernels · 7 datasets | test accuracy | No clear effect |
Closest papers by meaning
- Learning Kernels with Random Features (NeurIPS 2016 · similarity 0.86)
- DUAL: Learning Diverse Kernels for Aggregated Two-sample and Independence Testing (NeurIPS 2025 · similarity 0.86)
- Random Features for Kernel Approximation: A Survey on Algorithms, Theory, and Beyond (TPAMI 2021 · similarity 0.86)
- A Unified Theory of Diversity in Ensemble Learning (JMLR 2023 · similarity 0.86)
- From Kernels to Features: A Multi-Scale Adaptive Theory of Feature Learning (ICML 2025 · similarity 0.85)
- Class-Attribute Priors: Adapting Optimization to Heterogeneity and Fairness Objective (AAAI 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.
