Research Article
Rapid Identification of Asteraceae Plants with Improved RBF-ANN Classification Models Based on MOS Sensor E-Nose
Table 3
Comparison of three types of RBF-ANN with 12, 2, and 5 units.
| Sum of square error | Training | Testing | Classification accuracy | via 10-fold cross-validation |
| 12 units RBF-ANN | 0.939 | 0.320a | 100% | 2 units RBF-ANN | 0.083 | 0.029a | 100% | 5 units RBF-ANN | 0.522 | 0.207a | 100% |
|
|