Image Classification on MNIST Source (test)
69.67Accuracy (SVHN)Pro-RandConv
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Pro-RandConvBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 69.67 | 82.3 | 79.77 | 93.67 | 81.35 | |
| MetaCNNBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 66.5 | 88.27 | 70.66 | 89.64 | 78.76 | |
| Progressive (Same)Backbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization, Random convolution stacking=same weights2023.04 | 65.67 | 76.26 | 77.13 | 93.98 | 78.26 | |
| L2DBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 62.86 | 87.3 | 63.72 | 83.97 | 74.46 | |
| PDENBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 62.21 | 82.2 | 69.39 | 85.26 | 74.77 | |
| RandConv*Backbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization, reproduced=true2023.04 | 61.66 | 84.53 | 67.87 | 85.31 | 74.84 | |
| Progressive (Diff)Backbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization, Random convolution stacking=different weights2023.04 | 60.73 | 78.47 | 71.46 | 88.2 | 74.72 | |
| ME-ADABackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 42.56 | 63.27 | 50.39 | 81.04 | 59.32 | |
| M-ADABackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 42.55 | 67.94 | 48.95 | 78.53 | 59.49 | |
| ADABackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 35.51 | 60.41 | 45.32 | 77.26 | 54.62 | |
| JiGenBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 33.8 | 57.8 | 43.79 | 77.15 | 53.14 | |
| Baseline (ERM)Backbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 32.52 | 54.92 | 42.34 | 78.21 | 52 | |
| d-SNEBackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 26.22 | 50.98 | 37.83 | 93.16 | 52.05 | |
| CCSABackbone=LeNet, Source Domain=MNIST, Training Samples=10,000, Evaluation Protocol=Single domain generalization2023.04 | 25.89 | 49.29 | 37.31 | 83.72 | 49.05 |