AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Barriers for Learning in an Evolving World: Mathematical Understanding of Loss of Plasticity

Amir Joudaki, Giulia Lanzillotta, Mohammad Samragh, Seyed Iman Mirzadeh, Keivan Alizadeh, Thomas Hofmann, Mehrdad Farajtabar, Fartash Faghri

ICLR 2026 · Optimization › Implicit bias and generalization · approach: Edge-of-stability dynamics · 3 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
inject Gaussian gradient noise (relative magnitude 0.01) after cloningSGD on cloned MLP · Cloning MLP experimentescape from cloning manifold: single step noise 0.01 relative to gradient normImproved
add dropout to cloned network (ablation)cloned MLP · Cloning MLP experimentescape from cloning manifoldImproved
add noise perturbation to cloned ViTcloned ViT · Cloning ViT experimentdistance from manifoldNo clear effect
add normalization layers (BN/LN) (ablation)MLP/CNN/ResNet/ViT · Continual Tiny ImageNet (40 5-class tasks)effective rankImproved
add normalization to already cloned network (ablation)cloned network · Cloning experimentsbreaking clonesNo clear effect
switch from SGD to Continual Backpropagation in second halfSGD · Bit-flipping (5M samples)representation rank and online lossImproved
add dropout in continual bit-flipping (ablation)network · Bit-flippingplasticityMixed

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