AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Optimization, Generalization and Differential Privacy Bounds for Gradient Descent on Kolmogorov–Arnold Networks

Puyu Wang, Junyu Zhou, Philipp Liznerski, Marius Kloft

ICML 2026 · Optimization › Implicit bias and generalization · approach: Stability-based generalization analysis · 0 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
increase KAN width m from 8 to 512two-layer KAN · Multi-interval syntheticempirical NTK margin: 0.0400 -> 0.1027Improved
use KAN instead of parameter-matched ReLU networktwo-layer network · Multi-interval synthetic (m=512)empirical NTK margin: 0.0197 -> 0.1027Improved
use KAN instead of parameter-matched ReLU networktwo-layer network · XOR-gap synthetic (m=512)empirical NTK margin: 0.0284 -> 0.0599Improved
use KAN instead of parameter-matched ReLU networktwo-layer network · Checkerboard-gap synthetic (m=512)empirical NTK margin: 0.0111 -> 0.0538Improved
use KAN instead of ReLU at small widthtwo-layer network · XOR-gap synthetic (m=8)empirical NTK margin: 0.0226 -> 0.0243No clear effect
increase ReLU width from 64 to 512parameter-matched ReLU network · XOR-gap syntheticempirical NTK margin: 0.0285 -> 0.0284No clear effect
increase width m beyond about 24 (GD)two-layer KAN trained with GD · synthetic logistic data and MNISTtrain and test accuracyNo clear effect
increase GD iterations T beyond about 256two-layer KAN trained with GD · synthetic logistic data and MNISTtest accuracyNo clear effect
increase width or iterations beyond about 64 under DP (epsilon=2)two-layer KAN trained with DP-GD · synthetic logistic dataprivate test accuracyHurt

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