AI Paper Atlas

Research map / Continual learning / Continual fine-tuning of LLMs

PeriUn: Enhancing Unlearning by Selectively Forgetting Peripheral Samples

Hee Bin Yoo, Dong-Sig Han, Jaein Kim, Byoung-Tak Zhang

AAAI 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Machine unlearning by selective forgetting · 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
unlearn only low-confidence peripheral samples of the forget setSaliency Unlearn · TinyImageNet / VGG16-bnToW: 74.01 -> 96.05Improved
unlearn only peripheral samplesSaliency Unlearn · TinyImageNet / ResNet18ToW: 89.39 -> 97.55Improved
unlearn only peripheral samplesSaliency Unlearn · CIFAR100 / ResNet50ToW: 94.58 -> 97.78Improved
unlearn only peripheral samplesSaliency Unlearn · CIFAR10 / ResNet18ToW: 96.77 -> 97.92Improved
unlearn only peripheral samplesSaliency Unlearn · CIFAR10 / ResNet18confidence MIA gap: 8.53 -> 2.73Improved
unlearn only peripheral samplesSaliency Unlearn · TinyImageNet / ResNet18confidence MIA gap: 33.18 -> 12.34Improved
reduce selection ratio from 1.0 to 0.5 (ablation)SU+PeriUn · TinyImageNet / VGG16-bnToW: 74.01 -> 96.05Improved
too small selection ratio alpha=0.1 (ablation)SU+PeriUn · TinyImageNet / ResNet18ToW: 89.39 -> 80.59Hurt
too small selection ratio alpha=0.1 (ablation)SU+PeriUn · CIFAR100 / ResNet50ToW: 94.58 -> 84.14Hurt
apply PeriUn to Random LabelRL · CIFAR100 / ResNet50ToW: 59.33 -> 95.65Improved
apply PeriUn to GAGA · TinyImageNet / ResNet18ToW: 64.14 -> 64.11No clear effect
apply PeriUn to GAGDGAGD · CIFAR10 / ResNet18ToW: 97.12 -> 96.64Mixed

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