Image Classification Robustness on MNIST (test) (Noise Sensitivity) benchmark leaderboard · SOTA2 Research

Image Classification Robustness on MNIST (test) (Noise Sensitivity)

99.3Accuracy (Symmetric Noise η=0.2)SPR

Evaluation Results

Method	Links
SPR2022.03		99.3	—	99.2	99.2	98.7	99.3	99.2	99.2	—	—	—	—	—	—
SR2022.03		99.2	—	99.2	98.9	98	99.3	99.2	99.2	—	—	—	—	—	—
APL2022.03		99.1	—	98.4	95.7	73	98.9	96.9	91.5	—	—	—	—	—	—
SL2019.08		99.02	99.32	98.97	97.4	65.02	99.18	98.85	98	—	—	—	—	—	—
GCE2022.03		98.9	—	97.2	81.5	34	96.7	89.1	81.5	—	—	—	—	—	—
SCE2022.03		98.9	—	97.4	88.8	48.8	98	93.7	85.4	—	—	—	—	—	—
CNLCU-S2021.06		98.82	—	98.31	—	—	98.93	—	97.67	98.86	97.71	99.09	98.02	98.77	97.78
D2L2019.08		98.8	99.27	98.49	93.61	48.57	98.71	97.77	93.32	—	—	—	—	—	—
CNLCU-H2021.06		98.7	—	98.24	—	—	99.01	—	98.01	98.44	97.37	98.89	97.92	98.74	97.42
GCE2019.08		98.66	99.04	97.17	79.65	31.55	96.73	88.46	81.26	—	—	—	—	—	—
S2E2021.06		98.46	—	95.62	—	—	99.05	—	98.45	98.56	94.22	99.02	97.23	97.93	94.02
JoCor2021.06		98.42	—	98.04	—	—	98.05	—	94.55	98.01	96.85	98.45	96.98	98.62	96.07
NLNL2022.03		98.3	—	97.8	96.2	86.3	98.4	97.5	95.8	—	—	—	—	—	—
Co-teaching2021.06		97.53	—	95.62	—	—	98.25	—	95.08	96.05	94.16	98.05	96.18	97.96	95.02
MentorNet2021.06		95.04	—	92.08	—	—	96.32	—	90.86	93.19	90.93	96.42	93.28	94.65	90.11
Forward2019.08		94.85	99.03	86.02	69.77	49.72	97.31	96.25	95.72	—	—	—	—	—	—
SIGUA2021.06		92.31	—	91.88	—	—	93.96	—	62.59	93.77	86.22	94.92	83.46	92.9	86.34
FL2022.03		91.7	—	74.5	50.4	22.7	94.3	89.1	82.1	—	—	—	—	—	—
CE2022.03		91.6	—	74	49.4	22.7	94.6	88.8	82.3	—	—	—	—	—	—
LSR2019.08		89.56	99.28	69.56	45.01	21.28	94.18	88.39	81.09	—	—	—	—	—	—
Bootstrap2019.08		88.72	99.08	69.56	47.06	22.6	93.31	87.87	80.46	—	—	—	—	—	—
CE2019.08		88.71	99.02	69.97	46.54	21.77	93.14	87.91	81.1	—	—	—	—	—	—