Machine Unlearning on Tiny-Imagenet Random Forget 50%, γ=0 (test) benchmark leaderboard · SOTA2 Research

Machine Unlearning on Tiny-Imagenet Random Forget 50%, γ=0 (test)

4.32MIAl1-sparse

Evaluation Results

Method	Links
l1-sparse2026.01		4.32	95.68	95.48	55.45	32.37	—	—	—
IU2026.01		4.34	95.65	95.5	55.65	32.41	—	—	—
BS2026.01		4.34	95.66	95.47	55.03	32.26	—	—	—
BE2026.01		4.36	95.64	95.47	55.29	32.32	—	—	—
FT2026.01		4.41	95.59	99.99	54.71	31.02	—	—	—
GA2026.01		4.42	95.58	95.49	55.45	32.32	—	—	—
FalW2026.01		57.24	41.76	99.99	40.53	0.85	—	—	—
Retrain2026.01		59.4	40.6	99.99	40.63	0	—	—	—
SFRon2026.01		61.1	38.89	99.99	40.45	0.9	—	—	—
SalUn2026.01		61.51	38.49	97.21	36.23	2.85	—	—	—
RL2026.01		67.84	32.16	96.91	37.33	5.82	—	—	—
Retraining2026.04		89.1	—	—	—	0	73.5	31.4	34.7
RL2026.04		89.9	—	—	—	8.6	54.5	40.6	40.2
SALUN2026.04		91.4	—	—	—	13.4	31.2	27.9	29.3
SCRUB2026.04		94.8	—	—	—	9.2	57.1	40.8	40.3
IEU w/GA+Noisy2026.04		95	—	—	—	3.4	79.4	33.2	35
FT2026.04		95.1	—	—	—	4.4	72	38.7	37.6
IEU w/Noisy2026.04		95.1	—	—	—	3.4	79.2	33.2	34.8
IEU w/GA2026.04		95.3	—	—	—	4.5	82.6	34.1	34.7