AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

The Surprising Simplicity of the Early-Time Learning Dynamics of Neural Networks

Wei Hu, Lechao Xiao, Ben Adlam, Jeffrey Pennington

NeurIPS 2020 · Architectures and efficiency › RNNs and normalization · approach: Theory of depth and width · 18 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
Approximate early training of the two-layer network with gradient descent on its corresponding linear modelTwo-layer fully connected network · Synthetic Gaussian dataprediction discrepancyNo clear effect
Continue training the network after the linear model reaches its optimumTwo-layer fully connected network · Synthetic Gaussian datatraining progress: 1,000 stepsMixed
Increase the input dimensionTwo-layer network and corresponding linear model · Synthetic Gaussian dataprediction-discrepancy MSEImproved
Include the norm-dependent feature in the corresponding linear model (ablation)Two-layer ReLU network · Synthetic norm-dependent targetapproximation to network predictionsImproved
Use a naive linear model with the norm-feature coefficients set to zero (ablation)Corresponding linear model for a two-layer ReLU network · Synthetic norm-dependent targetapproximation to network predictionsHurt
Approximate early training of the network with its corresponding linear modelFour-hidden-layer fully connected network and CNN · Downsampled CIFAR-10 cats vs. horsestest loss: 1,000 stepsNo clear effect
Continue training beyond the early linear-learning phaseFour-hidden-layer fully connected network and CNN · Downsampled CIFAR-10 cats vs. horsestest loss in the complement of the input span: 1,000 stepsImproved
Approximate early training with the corresponding linear modelFour-hidden-layer CNN · Full-size CIFAR-10 binary classificationearly-time agreementMixed
Bound the initialization-kernel discrepancy in spectral norm rather than relying on Frobenius normTwo-layer network NTK analysis · Synthetic Gaussian datakernel-discrepancy scaling with dimension: d^{-1.263} vs. d^{-0.718}Improved

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