Research Article

A Novel Feature Selection Scheme and a Diversified-Input SVM-Based Classifier for Sensor Fault Classification

Table 9

Accuracies (%) of classifiers obtained in experiment 4.

ClassifierNormalErraticDriftHardoverSpikeStuckTotal

DI-SVM98.33100.00100.00100.0098.8899.4499.44
SVM93.6199.1699.7299.7297.2297.7797.87
NN98.6198.3398.0593.3388.6190.2794.53
KNN83.3392.7774.7283.3398.0583.3385.92