AI Paper Atlas

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

Forgetting Knowledge Localization and Isolation for Continual Forgetting of Pre-trained Vision Models

Zhiwen Yang, Jiehua Zhang, Chenggang Yan, Yuhan Gao, Zongpeng Li, Xichun Sheng, Liang Li

AAAI 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 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
add FKPI parameter isolation alone (ablation)continual forgetting baseline · ImageNet100H: 84.79 -> 86.16Improved
add FKLL layer localization alone (ablation)continual forgetting baseline · ImageNet100H: 84.79 -> 83.06Hurt
add all four components (FKLL, FKPI, LFKIL, LAlign) (ablation)continual forgetting baseline · ImageNet100H: 84.79 -> 87.01Improved
add FKLL on top of FKPI (ablation)FKPI · ImageNet100H: 86.16 -> 85.23Hurt
adaptive layer localization instead of random layers (ablation)FKLL · ImageNet100H: 81.37 -> 87.01Improved
adaptive per-task layer selection instead of fixed top-12 layers (ablation)FKLL · ImageNet100H: 85.01 -> 87.01Improved
SVD-based sensitivity initialization of LoRA instead of zero init (ablation)FKPI · ImageNet100H: 82.98 -> 87.01Improved
SVD-based initialization instead of standard LoRA initialization (ablation)FKPI · ImageNet100H: 72.98 -> 87.01Improved
increase LoRA rank (ablation)FKPI · ImageNet100HMixed
use our method instead of GS-LoRAcontinual forgetting · CASIA-Face100 (avg of 4 tasks)H: 70.0 -> 74.7Improved
use our method instead of GS-LoRAcontinual forgetting · ImageNet100 (avg of 4 tasks)H: 85.9 -> 87.0Improved

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