Classification Model Monitoring on MNIST
0.95Relative Error vs. RandomStratified
Evaluation Results
| Method | Links | |
|---|---|---|
| StratifiedSample size=100, Strata=PRED, Proposal=-2026.01 | 0.95 | |
| StratifiedSample size=1400, Strata=PRED, Proposal=-2026.01 | 0.99 | |
| Adaptive stratifiedSample size=100, Strata=PRED, Proposal=-2026.01 | 1.02 | |
| Adaptive stratifiedSample size=1400, Strata=PRED, Proposal=-2026.01 | 1.07 | |
| ImportanceSample size=1400, Strata=-, Proposal=q(x) ∝ score^0.252026.01 | 7.82 | |
| ImportanceSample size=100, Strata=-, Proposal=q(x) ∝ score^0.252026.01 | 7.84 | |
| SISSample size=100, Strata=PRED, Proposal=q(x) ∝ score^0.252026.01 | 8.97 | |
| SISSample size=1400, Strata=PRED, Proposal=q(x) ∝ score^0.252026.01 | 9.27 | |
| Adaptive ISSample size=100, Strata=PRED, Proposal=q(x) ∝ score^0.252026.01 | 10.31 | |
| Adaptive ISSample size=1400, Strata=PRED, Proposal=q(x) ∝ score^0.252026.01 | 12.82 |