The Compare of Microbial Electronic Tongue Data Based on Direct and Two-Stage Processing
In this paper, we used three working electrodes of platinum, gold, glassy carbon which constituted the microbial electronic tongue to identify microorganisms. The main purpose of this article is to classify sulfate-reducing bacteria and iron bacteria with the reference of broth culture. The accuracy of classification was compared based on directly processing (raw data processed by Back- Propagation Neural Network (BPNN), raw data processed by Partial LeastSquares (PLS) and raw data processed by Principal Components Analysis (PCA)) and two-stage processing (Principal Components Analysis (PCA) outputs processed by BPNN,PLS(score) outputs processed by BPNN and PLS(ypred) outputs processed by BPNN). The result showed that the performance of both PLS (ypred) and BPNN were optimum, which proved that the combine of linear and nonlinear model will get the best result for classification of bacterial.