AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Generalization Below the Edge of Stability: The Role of Data Geometry

Tongtong Liang, Alex Cloninger, Rahul Parhi, Yu-Xiang Wang

ICLR 2026 · Optimization › Implicit bias and generalization · approach: Stability-based generalization analysis · 1 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
change the input distribution from the uniform sphere to a union of 20 linestwo-layer ReLU network trained with GD · Synthetic quadratic teacher, d=50final true MSE: 1.0249 -> 0.07Improved
concentrate input mass away from the boundary (larger Beta-radial alpha, 0.1 to 5.0)two-layer ReLU network, d=5 · Beta-radial synthetic datalog-log error slope vs nImproved
increase ambient dimension d from 10 to 500 for a union of 20 linestwo-layer ReLU network · union of 20 lines (m=1)regression slope of error vs nNo clear effect
replace Gaussian inputs with MNIST inputstwo-layer ReLU network, n=30000 · Gaussian vs MNIST inputs (ground-truth teacher)time to memorize noisy labelsImproved
add low-dimensional structure to the data (sphere to mixture)two-layer ReLU network · neuron data activation rateshare of neurons firing on 10-40% of dataImproved
move data mass toward the sphere surface (shallow half-space depth)two-layer ReLU network · MNIST samplesprediction error vs half-space depthHurt

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