Research map / Learning theory and algorithms / Kernels and regression theory
FALKON: An Optimal Large Scale Kernel Method
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 FALKON instead of Prec. KRR | MillionSongs · MillionSongs | Relative error: 4.58\times10^{-3} -> 4.51\times10^{-3} | Helped |
| Use FALKON instead of Prec. KRR | MillionSongs · MillionSongs | Time (s): 289^\dagger -> 55 | Helped |
| Use FALKON instead of KRR | YELP · YELP | RMSE: 0.854 -> 0.833 | Helped |
| Use FALKON instead of KRR | YELP · YELP | Time (m): $500^\ddagger$ -> 20 | Helped |
| Use FALKON instead of Sparse Kernels | TIMIT · TIMIT | c-err: 30.9\% -> 32.3\% | Hurt |
| Use FALKON instead of EigenPro | SUSY · SUSY | c-err: 19.8\% -> \textbf{19.6\%} | Helped |
| Use FALKON instead of Deep Neural Network | SUSY · SUSY | AUC: 0.879 -> 0.877 | Hurt |
| Use FALKON instead of Deep Neural Network | HIGGS · HIGGS | AUC: 0.885 -> 0.833 | Hurt |
| Use FALKON instead of Inception-V4 | IMAGENET · IMAGENET | c-err: 20.0\% -> 20.7\% | Hurt |
| Increase Nyström centers to 5×10^4 (ablation) | FALKON · MillionSongs | MSE: 80.10 -> 79.20 | Helped |
| Increase Nyström centers to 5×10^4 (ablation) | FALKON · MillionSongs | Time (s): 55 -> 630 | Hurt |
| Increase Nyström centers to 10^5 (ablation) | FALKON · YELP | RMSE: 0.833 -> 0.828 | Helped |
Closest papers by meaning
- Fast kernel methods: Sobolev, physics-informed, and additive models (ICML 2026 · similarity 0.88)
- Kernel Methods Through the Roof: Handling Billions of Points Efficiently (NeurIPS 2020 · similarity 0.88)
- Distributed Second-Order Optimization using Kronecker-Factored Approximations (ICLR 2017 · similarity 0.86)
- gp2Scale: A Class of Compactly Supported Non-Stationary Kernels and Distributed Computing for Exact Gaussian Processes on 10 Million Data Points (ICML 2026 · similarity 0.86)
- Supervised Kernel Thinning (NeurIPS 2024 · similarity 0.85)
- Automated Kernel Discovery Towards Understanding High-dimensional Bayesian Optimization (NeurIPS 2026 · 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.