AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

On the Misalignment Between Data Learnability and Forgettability in Machine Unlearning

Zijie Pan, Zuobin Ying, Yajie Wang, Wanlei Zhou

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Low-rank compression and merging · 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
add Dual-Aware Training regularization (UGS variance and LFD penalty) to trainingCNN · CIFAR-10LFD: 0.31 -> 0.15 (-52%)Improved
add Dual-Aware Training to trainingTransformer · IMDBLFD: 0.28 -> 0.13 (-54%)Improved
add Dual-Aware Training to trainingCNN · MNISTLFD: 0.21 -> 0.10 (-52%)Improved
add Dual-Aware Training before unlearningCNN · CIFAR-10unlearning latency (s): 3.84 -> 1.04 (3.7×)Improved
add Dual-Aware Training before unlearningTransformer · IMDBunlearning latency (s): 0.90 -> 0.30 (3.0×)Improved
add Dual-Aware Training to trainingCNN · CIFAR-10accuracy (%): 93.1 -> 93.0No clear effect
add Dual-Aware Training to trainingCNN · MNISTaccuracy (%): 99.1 -> 99.1No clear effect
add Dual-Aware Training instead of post-hoc SISACNN · CIFAR-10LFD: 0.18 -> 0.15Improved
use SISA unlearning on top of vanilla training (training overhead)CNN · CIFAR-10training overhead (%): 45.0 vs 7.6Hurt
make regularization coefficients (lambda_ugs, lambda_lfd) very large (ablation)DAT · CIFAR-10test accuracyHurt
make regularization coefficients too small (ablation)DAT · CIFAR-10LFD and latencyHurt
increase Hutch++ sketch dimension k beyond 4 (ablation)DAT · CIFAR-10accuracy and LFD: k in {1, 2, 4, 8}No clear effect

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