Research map / Learning theory and algorithms / Kernels and regression theory
Sampling-based Nyström Approximation and Kernel Quadrature
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Mercer decomposition of kZ (kZ_s,mu / kZ_s,Y) instead of conventional SVD-based Nystrom kZ_s for choosing the low-rank kernel | kernel quadrature with non-i.i.d. landmark points Z · Periodic Sobolev spaces (Figures 1-2) | worst-case error of quadrature | Mixed |
| Replace SVD pseudo-inverse of kZ_s with Mercer-based kZ_s,mu/kZ_s,Y | kernel quadrature · Computational cost | runtime per trial | Hurt |
| Use i.i.d. landmark sampling with SVD-based Nystrom (improved bound) | conventional Nystrom kZ_s · theory | expected error bound | Helped |
| Remove inequality constraint (diagonal term) in kernel recombination with kZ_s | kZ_s kernel quadrature · Computational complexity | complexity | Helped |
Closest papers by meaning
- Recursive Sampling for the Nystrom Method (NeurIPS 2017 · similarity 0.93)
- Optimal Convergence Rates for Agnostic Nyström Kernel Learning (ICML 2023 · similarity 0.90)
- Fast DPP Sampling for Nystrom with Application to Kernel Methods (ICML 2016 · similarity 0.89)
- Positively Weighted Kernel Quadrature via Subsampling (NeurIPS 2022 · similarity 0.89)
- Improved guarantees and a multiple-descent curve for Column Subset Selection and the Nystrom method (NeurIPS 2020 · similarity 0.89)
- Entrywise error bounds for low-rank approximations of kernel matrices (NeurIPS 2024 · 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.