Official Journal of the Japan Wood Research Society
Table 2 Accuracy of regular CNN in prediction of specimen’s ORP and ρ
Splitting method for regular CNN | Group | ORP (°) | Density (g/cm3) | ||
|---|---|---|---|---|---|
R2 | RMSE | R2 | RMSE | ||
Method #1 | Train | 0.991 (0.004) | 2.4 (0.6) | 0.674 (0.102) | 0.034 (0.005) |
Test | 0.972 (0.008) | 4.4 (0.4) | 0.357 (0.061) | 0.048 (0.003) | |
Method #2 | TrainA2,B1,B2 | 0.990 | 2.7 | 0.809 | 0.021 |
TestA1 | 0.892 | 8.9 | − 469.166 | 0.112 | |
TrainA1,B1,B2 | 0.986 | 3.0 | 0.677 | 0.039 | |
TestA2 | 0.927 | 6.9 | − 4.211 | 0.049 | |
TrainA1,A2,B2 | 0.986 | 3.5 | 0.073 | 0.056 | |
TestB1 | 0.973 | 3.4 | − 10.321 | 0.045 | |
TrainA1,A2,B1 | 0.986 | 3.1 | 0.053 | 0.049 | |
TestB2 | 0.975 | 4.1 | − 5.815 | 0.085 | |