Image Classification on MNIST Corrupted benchmark leaderboard · SOTA2 Research

Image Classification on MNIST Corrupted

99.73Acc (0.01 Corruption)DROLT

Evaluation Results

Method	Links
DROLT2023.11		99.73	99.81	99.79	99.98	—
DRAUC-Da2023.11		99.56	99.91	99.94	99.96	—
CE2023.11		99.48	99.97	99.98	99.98	—
AUCMLoss2023.11		99.47	99.82	99.96	99.98	—
DROLT2023.11		99.46	99.79	99.88	99.96	—
ADVShift2023.11		99.46	99.83	99.97	99.98	—
DRAUC-Df2023.11		99.45	99.93	99.97	99.97	—
GLOT2023.11		99.39	99.91	99.94	99.97	—
CE2023.11		99.38	99.84	99.94	99.95	—
DRAUC-Da2023.11		99.34	99.86	99.94	99.97	—
AUCMLoss2023.11		99.26	99.87	99.92	99.96	—
WDRO2023.11		99.26	99.89	99.95	99.96	—
ADVShift2023.11		99.21	99.84	99.95	99.96	—
DRAUC-Df2023.11		99.19	99.9	99.94	99.96	—
WDRO2023.11		99.17	99.94	99.97	99.99	—
AUCDRO2023.11		99.12	99.82	99.92	99.95	—
FocalLoss2023.11		98.9	99.85	99.95	99.97	—
GLOT2023.11		98.8	99.88	99.96	99.98	—
AUCDRO2023.11		98.65	99.87	99.89	99.95	—
FocalLoss2023.11		67.05	98.64	99.39	99.73	—
CE2025.01		—	—	—	—	88.33
CE2025.01		—	—	—	—	85.5
CE2025.01		—	—	—	—	67.29
DICE2025.01		—	—	—	—	88.15
DICE2025.01		—	—	—	—	85.16
DICE2025.01		—	—	—	—	67.3
HEM-2025.01		—	—	—	—	84.57
HEM-2025.01		—	—	—	—	85.44
HEM-2025.01		—	—	—	—	68.12
LN2025.01		—	—	—	—	81.85
LN2025.01		—	—	—	—	83.18
LN2025.01		—	—	—	—	68.35
MM2025.01		—	—	—	—	87.93
MM2025.01		—	—	—	—	84.34
MM2025.01		—	—	—	—	67.75