AI Paper Atlas

Research map / Architectures and efficiency / CNN design

The Hessian perspective into the Nature of Convolutional Neural Networks

Sidak Pal Singh, Thomas Hofmann, Bernhard Schölkopf

ICML 2023 · Architectures and efficiency › CNN design · approach: Compact CNN architectures · 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
Use convolutional layers with local connectivity and weight sharingCNN Hessian rank · Deep CNNsHessian rank growth relative to parameter countImproved
Add weight sharing to a locally connected network (ablation)One-hidden-layer linear network with non-overlapping patches · Hessian rank boundsHessian rankImproved
Use local connectivity rather than full connectivity (ablation)One-hidden-layer network with non-overlapping patches · Hessian rank boundsHessian rankMixed
Increase the number of channelsCNN · Hessian rank versus channel countHessian rankMixed
Vary filter size (ablation)One-hidden-layer CNN · Hessian rank boundsHessian rankMixed
Use ReLU rather than linear activations (ablation)CNN · Empirical Hessian rank upper-bound verificationHessian rank bound validityNo clear effect
Use cross-entropy rather than MSE loss (ablation)CNN · Empirical Hessian rank upper-bound verificationHessian rank bound validityNo clear effect
Use larger networks with more channels (ablation)CNN · Hessian rank bound tightness across filter sizesHessian rank bound tightnessImproved

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