Machine Unlearning on SVHN 30% Random Forgetting VGG16-BN benchmark leaderboard · SOTA2 Research

Machine Unlearning on SVHN 30% Random Forgetting VGG16-BN

100Retention Accuracy (RA)FT

Evaluation Results

Method	Links
FT2026.04		100	4.84	0.23	3	3	0.28	0.05	0.48
Retrained (Baseline)2026.04		100	94.4	0.23	50	50	—	—	—
SalUn2026.04		99.91	0.81	0.04	0	0	0.19	0.06	0.55
CF-k2026.04		99.6	4.84	0.23	4	4	0.12	0.06	0.55
EU-k2026.04		99.6	4.85	0.23	4	4	0.25	0.07	0.56
Fisher2026.04		99.53	0.04	0	0	0	1.43	0.07	0.56
RELOAD2026.04		99.37	0.1	0.02	0	0	0.15	0.12	0.53
SCRUB2026.04		24.26	70.72	1.85	1	1	0.08	0.06	0.65
SSD2026.04		24.17	71.83	1.85	1	1	0.01	8.3	7.83
GA2026.04		16.05	79.62	0.25	5	1	0.01	0.06	0.66