Image Classification on MNIST 5000 train, 8000 (test)
96.74AccuracyRIEM-LA
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RIEM-LAPrior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=15, Batching Strategy=Full2023.06 | 96.74 | 0.115 | 0.0052 | 2.48 | 38.03 | |
| RIEM-LA (BATCHES)Prior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=15, Batch size=10002023.06 | 95.67 | 0.17 | 0.0072 | 5.4 | 22.4 | |
| LIN-RIEM-LAPrior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=15, Batching Strategy=Full2023.06 | 95.44 | 0.149 | 0.0068 | 0.66 | 39.4 | |
| LIN-RIEM-LA (BATCHES)Prior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=15, Batch size=10002023.06 | 95.14 | 0.167 | 0.0076 | 3.23 | 18.1 | |
| MAPPrior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=152023.06 | 95.02 | 0.167 | 0.0075 | 1.05 | 39.94 | |
| LIN-LAPrior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=152023.06 | 94.91 | 0.204 | 0.0087 | 6.3 | 39.3 | |
| VANILLA LAPrior Precision=Optimized, Monte Carlo samples=25, Backbone=CNN, Bin count (M)=152023.06 | 88.69 | 0.871 | 0.0393 | 42.11 | 50.52 |