AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Faster Directional Convergence of Linear Neural Networks under Spherically Symmetric Data

Dachao Lin, Ruoyu Sun, Zhihua Zhang

NeurIPS 2021 · Optimization › Implicit bias and generalization · approach: Low-rank and single-index recovery · 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
use spherically symmetric data instead of a finite positive-margin dataset for logistic regression directional convergencegradient descent on logistic regression · directional convergence rateconvergence rate of weight directionImproved
increase depth from N=1 or 2 to N>=3 layers in the linear networkdeep linear network under gradient flow · spherically symmetric dataeffect of weight norm on directional convergence speedMixed
scale up the data to increase expected norm c0second-phase convergence of deep linear network · spherically symmetric datasecond-phase convergence speedImproved
add the second phase (weight norm ascending) to training dynamicsdeep linear network gradient flow · x ~ U(S1) simulationdirectional convergence rateImproved
run SGD with batch size 1000 and learning rate 10^-3 instead of exact gradient flowlinear classifier and 4-layer linear network · x ~ U(S1) simulationmatch to theoretical boundsNo clear effect
track accuracy-related direction cos(theta) instead of loss alonedeep linear network training analysis · x ~ U(S1) simulationprogress during early trainingImproved

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