ResearchBenchmarksBackdoor Feature Unlearning on ImageNet Dr (retain)For AI assistantsMarkdownContent indexBackdoor Feature Unlearning on ImageNet Dr (retain)Follow67.34AccuracyFine-tune11.304825.852440.454.9476May 23, 2024Evaluation resultsMethodMethodLinksAccuracyFine-tune2024.0567.34Retrain2024.0567.05Ferrari2024.0565.74Baseline2024.0552.35FedCDP2024.0529.74FedRecovery2024.0513.46