Research Article
Mango Grading System Based on Optimized Convolutional Neural Network
Table 2
Influence of different parameters on error and accuracy of the data set.
| Experiment number | Learning rate | Momentum | Verification Top1 error | Test set accuracy | Loss |
| 1 | 0.008 | 0.5 | — | — | — | 2 | 0.7 | 0.800 | 28.4% | 8.70 | 3 | 0.9 | 0.800 | 23.5% | 10.18 | 4 | 0.005 | 0.5 | 0.353 | 60.0% | 1.39 | 5 | 0.7 | 0.529 | 33.3% | 1.72 | 6 | 0.9 | 0.765 | 20.5% | 1.93 | 7 | 0.001 | 0.5 | 0.176 | 80.0% | 0.55 | 8 | 0.7 | 0.168 | 86.6% | 0.48 | 9 | 0.9 | 0.114 | 88.7% | 0.43 |
|
|