Image Classification on CIFAR10 (test) (Acc, NLL, ECE, Brier Score)
43.61AccuracyBPC-fKL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| BPC-fKLImages per class (ipc)=10, Inference method=HMC2022.10 | 43.61 | 1.7198 | 15.38 | 0.7231 | |
| BPC-WImages per class (ipc)=10, Inference method=HMC2022.10 | 35.38 | 1.9369 | 14.57 | 0.803 | |
| BPC-rKLImages per class (ipc)=10, Inference method=HMC2022.10 | 33.34 | 1.9516 | 11.83 | 0.7988 | |
| HerdingImages per class (ipc)=10, Inference method=HMC2022.10 | 30 | 2.0343 | 12.09 | 0.8231 | |
| RandomImages per class (ipc)=10, Inference method=HMC2022.10 | 25.9 | 2.182 | 13.85 | 0.8595 | |
| K-centerImages per class (ipc)=10, Inference method=HMC2022.10 | 17.39 | 2.3934 | 13.6 | 0.9125 |