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

Image Classification on TinyImageNet (Adversarial Robustness Evaluation)

63Clean AccuracyBenign

Evaluation Results

Method	Links
Benign2026.05		63	1.05	0.03	0
Benign2026.05		61.78	0.92	0.06	0
SORA2026.05		54.59	22.84	18.73	14.31
ELLE2026.05		49.43	21.67	18.56	14.4
PGD-22026.05		49.4	22.34	19.43	15.13
N-FGSM2026.05		48.47	23.83	21.21	16.58
ELLE2026.05		48.42	21.26	18.41	13.87
GradAlign2026.05		48.19	22.3	19.47	15.12
PGD-22026.05		48.07	21.73	19.21	14.66
GradAlign2026.05		47	21.91	19.51	14.55
ZeroGrad2026.05		46.66	22.17	19.44	15.09
FGSM-RS2026.05		46.41	21.57	19.02	14.37
MultiGrad2026.05		46.24	22.69	20.26	15.78
FGSM-RS2026.05		45.92	17.59	0.05	0
SORA2026.05		45.83	22.86	20.45	15.72
NuAT2026.05		45.35	27.09	11.22	2.39
ZeroGrad2026.05		45.24	21.36	19.19	14.55
N-FGSM2026.05		45.19	22.36	20.16	15.68
AAER2026.05		45.11	22.45	20.29	15.85
MultiGrad2026.05		44.87	22.04	19.86	15.03
SORA2026.05		44.11	22.21	20.39	15.2
AAER2026.05		43.59	21.79	20.04	15.26
N-FGSM2026.05		43.4	21.88	20.17	15.25
ZeroGrad2026.05		43.13	59.34	0	0
FGSM2026.05		42.82	20.81	18.94	14.21
AAER2026.05		42.23	21.31	19.43	16.56
FGSM2026.05		39.1	38.16	5.75	4.3
FGSM2026.05		33.8	64.24	0.01	0
FGSM-RS2026.05		29.96	64.75	0	0
NuAT2026.05		26.88	16.91	16.44	11.95