Image Classification on MNIST, FashionMNIST, and CIFAR-10 Average
94.54Average AccuracyOracle
Evaluation results
| Method | Links | |
|---|---|---|
| OracleNORMpre=none, GEN=least-squares, NORMpost=none, NORMinf=none, AGG=max2025.03 | 94.54 | |
| IMPRINT (Ours)NORMpre=L2, GEN=k-means, NORMpost=L2, NORMinf=L2, AGG=max, k=202025.03 | 91.06 | |
| Qi et al. (2018)NORMpre=L2, GEN=mean, NORMpost=L2, NORMinf=L2, AGG=max2025.03 | 86.79 | |
| Janson et al. (2022)NORMpre=none, GEN=mean, NORMpost=none, NORMinf=none, AGG=1-nn2025.03 | 86.64 | |
| Hosoda et al. (2024)NORMpre=none, GEN=mean, NORMpost=quantile, NORMinf=none, AGG=max2025.03 | 82.9 |