AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Parallelizing Spectrally Regularized Kernel Algorithms

Nicole Mücke, Gilles Blanchard

JMLR 2018 · Learning theory and algorithms › Kernels and regression theory · approach: RKHS kernel methods · 31 citations (OpenAlex)

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 changedWhereMeasuredEffect
partition the data into m = n^alpha subsamples with alpha below 0.5 (low smoothness) (ablation)Distributed nu-method · Low smoothness simulationreconstruction error in HK-norm: alpha < 0.5No clear effect
use alpha above 0.6 partitioning (low smoothness) (ablation)Distributed nu-method · Low smoothness simulationreconstruction error in HK-norm: alpha > 0.6Hurt
apply overregularization (lower stopping index) in distributed learning (ablation)Distributed nu-method · Low smoothness simulationreconstruction error in HK-normHurt
use almost no regularization with partitioning as the only regularizer (ablation)Distributed nu-method · Low smoothness simulationreconstruction error in HK-normImproved
partition the data into m = n^alpha subsamples (high smoothness, r infinity) (ablation)Distributed gradient descent · High smoothness simulationreconstruction error in HK-normHurt
use severe underregularization with partitioning (high smoothness) (ablation)Distributed gradient descent · High smoothness simulationreconstruction error in HK-norm: roughly 0.25Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy