Image Classification on CIFAR-10 (Comprehensive Metrics including Calibration)
-0.3026LPPDDE
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| DEBackbone=ResNet-18, Parameters=11.2M, Number of ensemble members=82026.02 | -0.3026 | 89.95 | 89.38 | 0.1545 | 0.4507 | 99.34 | 1.61 | |
| SGHMCBackbone=ResNet-18, Parameters=11.2M, Number of ensemble members=82026.02 | -0.2908 | 91.04 | 91.03 | 0.1389 | 0.4942 | 99.49 | 1.27 | |
| cSGLDBackbone=ResNet-18, Parameters=11.2M, Number of ensemble members=82026.02 | -0.2794 | 91.32 | 91.31 | 0.1341 | 0.4952 | 99.53 | 1.16 | |
| SMILEBackbone=ResNet-18, Parameters=11.2M, Number of ensemble members=82026.02 | -0.2763 | 91.01 | 91 | 0.1344 | 0.4071 | 99.51 | 1.22 | |
| pSMILEBackbone=ResNet-18, Parameters=11.2M, Number of ensemble members=82026.02 | -0.2598 | 91.35 | 91.34 | 0.1264 | 0.39 | 99.54 | 1.13 |