In this initial RBF-ANN model, input layer contains 12 units from 12 MOS sensors.
Comparison of Original and Improved RBF-ANN Models.
Table 3 shows that the classification accuracies of three types of RBF-ANN models with 12, 2, and 5 units by 10-fold cross-validation are all 100%.
Then RBF-ANN was applied to establish the classification model.
Abbreviations TCM: Traditional Chinese medicine E-nose: Electronic nose MOS: Metal oxide semiconductor PCA: Principal component analysis BC: BestFirst + CfsSubsetEval RBF-ANN: Radial basis function artificial neural networks.