Comparison of Random Forest, Support Vector Machine and Back Propagation Neural Network for Electronic Tongue Data Classification: Application to the Recognition of Orange Beverage and Chinese Vinegar

Miao Liu,Mingjun Wang,Jun Wang,Duo Li
DOI: https://doi.org/10.1016/j.snb.2012.11.071
2013-01-01
Abstract:Random forest (RF) has been proposed on the basis of classification and regression trees (CART) with “ensemble learning” strategy by Breiman in 2001. In this paper, RF is introduced and investigated for electronic tongue (E-tongue) data processing. The experiments were designed for type and brand recognition of orange beverage and Chinese vinegar by an E-tongue with seven potentiometric sensors and an Ag/AgCl reference electrode. Principal component analysis (PCA) was used to visualize the distribution of total samples of each data set. Back propagation neural network (BPNN) and support vector machine (SVM), as comparative methods, were also employed to deal with four data sets. Five-fold cross-validation (CV) with twenty replications was applied during modeling and an external testing set was employed to validate the prediction performance of models. The average correct rates (CR) on CV sets of the four data sets performed by BPNN, SVM and RF were 86.68%, 66.45% and 99.07%, respectively. RF has been proved to outperform BPNN and SVM, and has some advantages in such cases, because it can deal with classification problems of unbalanced, multiclass and small sample data without data preprocessing procedures. These results suggest that RF may be a promising pattern recognition method for E-tongues.
What problem does this paper attempt to address?