AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearning

Chongyu Fan, Jiancheng Liu, Alfred Hero, Sijia Liu

ECCV 2024 · Learning theory and algorithms › Prediction and decision losses · approach: Data attribution and recourse · 37 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 on the bilevel-optimized worst-case forget set instead of a random forget setRetrain (exact unlearning) · CIFAR-10, ResNet-18UA: 5.28 -> 0.00 (10% forgetting)Improved
forget worst-case set instead of random setRetrain · CIFAR-10, ResNet-18TA: 94.38 -> 94.66 (10%); 94.04 -> 94.60 (20%)Improved
evaluate relabeling-based methods (RL, BE, BS, SalUn) on worst-case forget setRL, BE, BS, SalUn · CIFAR-10, 10% forgettingAvg. gap vs Retrain: SalUn 1.89 -> 24.51; BE 4.28 -> 34.70Hurt
evaluate relabeling-free methods on worst-case forget setFT, EUk, CFk, Scrub, l1-sparse · CIFAR-10, 10% forgettingAvg. gap vs Retrain: Scrub 9.91 -> 0.82; FT 2.00 -> 1.37Improved
worst-case forget set reduces evaluation variance versus random forget setScrub · CIFAR-10, 10% forgettingUA std: 12.42 ±19.82 -> 0.01 ±0.01Improved
train on complement of worst-case forget set as a coresetResNet-18 · CIFAR-10 / CIFAR-100test accuracyImproved
forget worst-case prompt set instead of random prompts with ESDStable Diffusion on UnlearnCanvas · UnlearnCanvas prompt-wise forgettingqualitative unlearning successHurt

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