Image Classification Training Time on CIFAR10 (train)
17.85Training Time (s)Uniform
Evaluation results
| Method | Links | ||
|---|---|---|---|
| UniformNoise=0.3 pair label noise2026.05 | 17.85 | — | |
| UniformNoise=0.4 symmetric label noise2026.05 | 17.9 | — | |
| UniformNoise=0.5 symmetric label noise2026.05 | 17.9 | — | |
| ADIW-KLNoise=0.3 pair label noise2026.05 | 38.48 | — | |
| ADIW-KLNoise=0.5 symmetric label noise2026.05 | 38.61 | — | |
| ADIW-KMNoise=0.5 symmetric label noise2026.05 | 38.73 | — | |
| ADIW-KLNoise=0.4 symmetric label noise2026.05 | 38.82 | — | |
| ADIW-WANoise=0.3 pair label noise2026.05 | 39.59 | — | |
| ADIW-WANoise=0.4 symmetric label noise2026.05 | 39.66 | — | |
| ADIW-WANoise=0.5 symmetric label noise2026.05 | 39.79 | — | |
| ADIW-KMNoise=0.4 symmetric label noise2026.05 | 49.71 | — | |
| ADIW-KMNoise=0.3 pair label noise2026.05 | 55.04 | — | |
| ADIW-LSNoise=0.4 symmetric label noise2026.05 | 67.94 | — | |
| ADIW-LSNoise=0.5 symmetric label noise2026.05 | 68.53 | — | |
| ADIW-LSNoise=0.3 pair label noise2026.05 | 69.59 | — | |
| A2CiD2n (number of workers)=64, Backbone=ResNet182023.06 | 90 | — | |
| DIWNoise=0.3 pair label noise2026.05 | 90.4 | — | |
| AR-SGDn (number of workers)=64, Backbone=ResNet182023.06 | 108 | — | |
| DIWNoise=0.4 symmetric label noise2026.05 | 109.55 | — | |
| DIWNoise=0.5 symmetric label noise2026.05 | 111.01 | — | |
| L2RNoise=0.3 pair label noise2026.05 | 119.66 | — | |
| L2RNoise=0.5 symmetric label noise2026.05 | 121.07 | — | |
| L2RNoise=0.4 symmetric label noise2026.05 | 121.38 | — | |
| MW-NetNoise=0.3 pair label noise2026.05 | 134.4 | — | |
| MW-NetNoise=0.5 symmetric label noise2026.05 | 135.15 | — | |
| MW-NetNoise=0.4 symmetric label noise2026.05 | 152.12 | — | |
| A2CiD2n (number of workers)=32, Backbone=ResNet182023.06 | 162 | — | |
| AR-SGDn (number of workers)=32, Backbone=ResNet182023.06 | 192 | — | |
| A2CiD2n (number of workers)=16, Backbone=ResNet182023.06 | 312 | — | |
| AR-SGDn (number of workers)=16, Backbone=ResNet182023.06 | 396 | — | |
| A2CiD2n (number of workers)=8, Backbone=ResNet182023.06 | 630 | — | |
| AR-SGDn (number of workers)=8, Backbone=ResNet182023.06 | 666 | — | |
| CCEEpochs=2002021.06 | 825.68 | — | |
| BAREEpochs=2002021.06 | 930.78 | — | |
| A2CiD2n (number of workers)=4, Backbone=ResNet182023.06 | 1,254 | — | |
| CLEpochs=2002021.06 | 1,254.3 | — | |
| AR-SGDn (number of workers)=4, Backbone=ResNet182023.06 | 1,314 | — | |
| COTEpochs=2002021.06 | 1,687.9 | — | |
| COT+Epochs=2002021.06 | 1,790.57 | — | |
| MREpochs=2002021.06 | 8,130.87 | — | |
| MNEpochs=2002021.06 | 8,891.6 | — | |
| Full DPBackbone=ResNet-18, epsilon=0.22026.01 | 80,110 | — | |
| Full DPBackbone=ResNet-18, epsilon=0.42026.01 | 81,511 | — | |
| Full DPBackbone=ResNet-18, epsilon=0.52026.01 | 82,018 | — | |
| Full DPBackbone=ResNet-18, epsilon=0.32026.01 | 82,157 | — | |
| Sensitive DPBackbone=ResNet-18, epsilon=0.42026.01 | 82,443 | — | |
| Time Varying DPBackbone=ResNet-18, epsilon=0.32026.01 | 83,151 | — | |
| Sensitive DPBackbone=ResNet-18, epsilon=0.22026.01 | 83,793 | — | |
| Sensitive DPBackbone=ResNet-18, epsilon=0.52026.01 | 84,008 | — | |
| Sensitive DPBackbone=ResNet-18, epsilon=0.32026.01 | 86,363 | — | |
| Time Varying DPBackbone=ResNet-18, epsilon=0.22026.01 | 86,423 | — | |
| Time Varying DPBackbone=ResNet-18, epsilon=0.42026.01 | 87,562 | — | |
| Time Varying DPBackbone=ResNet-18, epsilon=0.52026.01 | 89,110 | — | |
| DPA LDPBackbone=ResNet-18, epsilon=0.32026.01 | 89,569 | — | |
| DPA LDPBackbone=ResNet-18, epsilon=0.42026.01 | 90,524 | — | |
| DPA LDPBackbone=ResNet-18, epsilon=0.22026.01 | 91,687 | — | |
| AdapLDPBackbone=ResNet-18, epsilon=0.32026.01 | 92,042 | — | |
| DPA LDPBackbone=ResNet-18, epsilon=0.52026.01 | 92,272 | — | |
| LaDP-FLBackbone=ResNet-18, epsilon=0.32026.01 | 93,455 | — | |
| LaDP-FLBackbone=ResNet-18, epsilon=0.22026.01 | 93,775 | — | |
| AdapLDPBackbone=ResNet-18, epsilon=0.22026.01 | 94,773 | — | |
| LaDP-FLBackbone=ResNet-18, epsilon=0.42026.01 | 94,985 | — | |
| AdapLDPBackbone=ResNet-18, epsilon=0.52026.01 | 95,647 | — | |
| LaDP-FLBackbone=ResNet-18, epsilon=0.52026.01 | 96,008 | — | |
| AdapLDPBackbone=ResNet-18, epsilon=0.42026.01 | 96,111 | — | |
| AdapLDPBackbone=ResNet-182026.01 | — | -0.08 | |
| DPA LDPBackbone=ResNet-182026.01 | — | 3.9 | |
| Full DPBackbone=ResNet-182026.01 | — | 16.1 | |
| Sensitive DPBackbone=ResNet-182026.01 | — | 12.41 | |
| Time Varying DPBackbone=ResNet-182026.01 | — | 9.28 |