Adversarial Robustness Evaluation on TinyImageNet (test) benchmark leaderboard · SOTA2 Research

Adversarial Robustness Evaluation on TinyImageNet (test)

63.27Standard AccuracyNatural Training

Evaluation Results

Method	Links
Natural Training2026.01		63.27	0.01	0.01	0	0
FGSM-CKPT2026.01		57.93	11.68	10.49	9.89	9.07
NuAT2026.01		53.97	18.25	17.52	17.06	13.5
FGSM-CKPT + QUB-decreasing2026.01		52.11	19.13	18.33	17.83	15.19
GAT2026.01		49.75	17.71	17.3	16.93	12.62
FGSM-RS2026.01		48.14	20.54	19.98	19.67	16.39
TRADES2026.01		46.7	21.76	21.56	21.38	16.33
FGSM-UAP2026.01		46.25	21.86	21.53	21.36	17.26
FGSM-GA2026.01		44.98	21.25	20.75	20.52	16.46
N-FGSM2026.01		44.96	21.2	20.8	20.52	16.7
ELLE-A2026.01		43.95	17.62	16.91	16.34	13.42
FGSM-PGI(MEP) + QUB-decreasing2026.01		43.32	23.79	23.4	23.38	18.03
PGD AT2026.01		43.15	21.93	21.67	21.44	17.44
FGSM-PGI(MEP)2026.01		42.72	23.63	23.35	23.07	17.58
FGSM-UAP + QUB-decreasing2026.01		42.45	22.49	22.2	21.94	17.71
N-FGSM + QUB-decreasing2026.01		41.97	21.67	21.31	21.1	17.18
ELLE-A + QUB-decreasing2026.01		41.97	21.67	21.31	21.1	17.18
PGD AT + QUB-decreasing2026.01		41	22.94	22.75	22.61	18.38
FGSM-GA + QUB-decreasing2026.01		38.39	21.87	21.57	21.33	17.21
Free-AT2026.01		36.32	16.22	16.04	15.91	12.03
Free-AT + QUB-decreasing2026.01		31.34	16.59	16.47	16.38	12.51
FGSM-RS + QUB-decreasing2026.01		29.47	16.02	15.89	15.77	11.73