Research Article
Convolutional Neural Network-Based Discriminator for Outlier Detection
Table 13
Average test accuracy on CIFAR-10 over the last ten epochs.
| | Method | Flipping rate | | Symmetry-20% | Symmetry-50% | Pair-45% |
| | Standard | 76.25 (±0.28) | 48.87 (±0.52) | 49.50 (±0.42) | | Bootstrap | 77.01 (±0.29) | 50.66 (±0.56) | 50.05 (±0.30) | | S-model | 76.84 (±0.66) | 46.15 (±0.76) | 48.21 (±0.55) | | F-correction | 84.55 (±0.16) | 59.83 (±0.17) | 6.61 (±1.12) | | Decoupling | 80.44 (±0.05) | 51.49 (±0.08) | 48.80 (±0.04) | | MentorNet | 80.76 (±0.36) | 71.10 (±0.48) | 58.14 (±0.38) | | Coteaching | 82.32 (±0.07) | 74.02 (±0.04) | 72.62 (±0.15) |
| | EBF | 85.58 (±0.58) | 74.30 (±1.26) | 59.17 (±1.91) | | Discriminator (100S) | 78.09 (±0.90) | 70.84 (±0.81) | 74.24 (±1.31) | | Discriminator (200S) | 79.79 (±0.61) | 72.44 (±1.93) | 76.33 (±1.15) | | Discriminator (400S) | 84.72 (±0.53) | 79.05 (±0.72) | 80.57 (±1.21) |
|
|
