AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Gradient sparsity regularization for training unlearning-compatible models

Sobhan Hemati, Soufiane Lamghari, Masoud Asgharian, Xu Li, Hongliang Li

ECCV 2026 · Optimization › Implicit bias and generalization · approach: Stability-based generalization analysis · 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
Train with gradient l1 regularization before unlearningFT unlearning · CIFAR-10 class forgetting (ResNet-18)Avg. Gap: 22.61 -> 0.81Improved
Train with gradient l1 regularization before unlearningIU unlearning · CIFAR-10 class forgetting (ResNet-18)Avg. Gap: 40.61 -> 5.76Improved
Train with gradient l1 regularization before unlearningWS unlearning · CIFAR-10 random forgetting (ResNet-18)Avg. Gap: 7.87 -> 2.06Improved
Train with gradient l1 regularization before unlearningFF unlearning · CIFAR-10 random forgetting (ResNet-18)Avg. Gap: 10.83 -> 10.59No clear effect
Train with gradient l1 regularization before unlearningFT unlearning · CIFAR-10 random forgetting (VGG16)Avg. Gap: 3.54 -> 4.14Hurt
Train with gradient l1 regularization before unlearningGA, IU, FF, FT · CIFAR-100 random forgetting (ResNet-18)Avg. Gap: FT 15.08 -> 14.34No clear effect
Train with gradient l1 regularization before unlearningGA unlearning · ImageNet class forgetting (ResNet-18)Avg. Gap: 9.53 -> 5.72Improved
Train with gradient l1 regularization before unlearningFT unlearning · ImageNet class forgetting (ResNet-18)Avg. Gap: 17.20 -> 16.81No clear effect
Train with gradient l1 regularization before unlearningIU unlearning · SVHN class forgetting (ViT-small)Avg. Gap: 42.36 -> 29.1Improved
Add gradient sparsity regularization under backdoor poisoningMobileNetV2 · Poisoned CIFAR-10 (3%)poison test accuracy: 86.62 -> 24.63Improved
Add gradient sparsity regularization under backdoor poisoningMobileNetV2 · Poisoned CIFAR-10 (3%)clean test accuracy: 68.19 -> 68.1No clear effect
Use gradient l1 norm instead of l2 norm regularizer (ablation)FT unlearning · CIFAR-10 class forgettingAvg. GapImproved

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