# Class-wise Machine Unlearning on CIFAR-20 veg (test) Source: [Class-wise Machine Unlearning on CIFAR-20 veg (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:34:58.949Z Evaluation of class-wise machine unlearning effectiveness on the test split of the CIFAR-20 veg dataset, focusing on utility (Ar, Af) and membership inference attack (MIA) metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Class-wise Machine Unlearning | | Dataset (including any stated split) | CIFAR-20 veg (test) | | Metric | Ar (Accuracy Retention) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Ar (Accuracy Retention)](): Higher is better; unit: PERCENT. - [Af (Accuracy Forgetting)](): Higher is better; unit: PERCENT. - [MIA Success Rate](): Higher is better; unit: PERCENT. ## Recorded results | Method | Ar (Accuracy Retention) | Result date | Report paper | | --- | --- | --- | --- | | baseline | 95.88 | Not recorded | 6997ea6609c33900fcd76493 | | MaGA | 95.64 | Not recorded | 6997ea6609c33900fcd76493 | | SSD | 95.61 | Not recorded | 6997ea6609c33900fcd76493 | | retrain | 94.95 | Not recorded | 6997ea6609c33900fcd76493 | | UNSIR | 93.75 | Not recorded | 6997ea6609c33900fcd76493 | | AMNC | 93.31 | Not recorded | 6997ea6609c33900fcd76493 | | FT | 88.52 | Not recorded | 6997ea6609c33900fcd76493 | | SSD | 82.64 | Not recorded | 6997ea6609c33900fcd76493 | | baseline | 82.59 | Not recorded | 6997ea6609c33900fcd76493 | | MaGA | 82.53 | Not recorded | 6997ea6609c33900fcd76493 | | retrain | 82.24 | Not recorded | 6997ea6609c33900fcd76493 | | AMNC | 81.66 | Not recorded | 6997ea6609c33900fcd76493 | | UNSIR | 81.23 | Not recorded | 6997ea6609c33900fcd76493 | | FT | 72.56 | Not recorded | 6997ea6609c33900fcd76493 | ### baseline Ar (Accuracy Retention): 95.88. Result ID: 699b058f5698665ec497e49a. [Reporting paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### MaGA Ar (Accuracy Retention): 95.64. Result ID: 699b058f5698665ec497e4a0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### SSD Ar (Accuracy Retention): 95.61. Result ID: 699b058f5698665ec497e49f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### retrain Ar (Accuracy Retention): 94.95. Result ID: 699b058f5698665ec497e49b. [Reporting paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### UNSIR Ar (Accuracy Retention): 93.75. Result ID: 699b058f5698665ec497e49d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### AMNC Ar (Accuracy Retention): 93.31. Result ID: 699b058f5698665ec497e49e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### FT Ar (Accuracy Retention): 88.52. Result ID: 699b058f5698665ec497e49c. [Reporting paper]() Recorded evaluation configuration: {"model":"ViT"}. This record does not assert independent reproduction. ### SSD Ar (Accuracy Retention): 82.64. Result ID: 699b058f5698665ec497e498. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### baseline Ar (Accuracy Retention): 82.59. Result ID: 699b058f5698665ec497e493. [Reporting paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### MaGA Ar (Accuracy Retention): 82.53. Result ID: 699b058f5698665ec497e499. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### retrain Ar (Accuracy Retention): 82.24. Result ID: 699b058f5698665ec497e494. [Reporting paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### AMNC Ar (Accuracy Retention): 81.66. Result ID: 699b058f5698665ec497e497. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### UNSIR Ar (Accuracy Retention): 81.23. Result ID: 699b058f5698665ec497e496. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. ### FT Ar (Accuracy Retention): 72.56. Result ID: 699b058f5698665ec497e495. [Reporting paper]() Recorded evaluation configuration: {"model":"RN18"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.