AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

An Information Theoretic Evaluation Metric for Strong Unlearning

Dongjae Jeon, Wonje Jeung, Taeheon Kim, Albert No, Jonghyun Choi

AAAI 2026 · Adversarial robustness and security › Black-box and transfer attacks · approach: Data-free substitute training · 18 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
evaluate unlearning with IDI in addition to black-box MIA/JSDhead distillation (HD) on CIFAR-10 ResNet-18 · single-class forgettingblack-box vs white-box efficacy: HD MIA 2.05 but IDI 1.000Improved
retrain a new head with 2% of data on frozen unlearned encoders (ablation)RL, SALUN, Bad-T unlearned encoders · CIFAR-10 ResNet-18, forget test accuracyforget test accuracy: Retrain no more than 41% vs over 82% for Bad-T, SALUN, RLHurt
use COLA (supervised contrastive collapse on retain set, then cross-entropy alignment) instead of RL/SALUNclass-wise forgetting · CIFAR-10 ResNet-18IDI: 0.830 (RL) -> 0.010Improved
use COLA instead of SCRUBclass-wise forgetting · ImageNet-1K ResNet-50IDI: 0.517 -> 0.040Improved
use COLA instead of Retrainclass-wise forgetting · ImageNet-1K ResNet-50runtime (min): 2661.90 -> 171.44Improved
use COLA instead of faster methods like HD/EU-5/FTclass-wise forgetting · CIFAR-10 ResNet-18runtime (min): 1.54 (EU-5) -> 4.91Hurt
use COLA+ for random data forgetting instead of SALUNrandom data forgetting · CIFAR-10 ResNet-18IDI: 0.861 -> 0.024Improved
use IDI instead of white-box MIA to evaluate unlearningevaluation metric · CIFAR-100 single-class forgetting ResNet-18metric stability: Random model MIA 53.20 vs Retrain 52.87; IDI -2.955Improved
forget set data close to Retrain MIA does not imply low IDI (CF-5)CF-5 · ImageNet-1K five classes ResNet-50MIA vs IDI: MIA 10.21 vs 9.41 Retrain, IDI 0.701Mixed

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