Computer aided test and diagnosis model for exceptional signals detection and classification by using WT-SVM

In order to detect and diagnosis the exceptional signals, this paper presents a new computer aided test and diagnosis (CAT/CAD) model based on wavelet transform (WT) and support vector machine (SVM). The architecture of the CAT and CAD model and experimental feature extraction and pattern recognition by WT-SVM system are presented. The features of special frequency segment of the signal picked up by the method of wavelet decomposition are used as the inputs of SVM. The analysis of the experimental data proves that the model proposed is efficient and simple to recognize exceptional signals. The experimental result indicated that the error rate of SVM is lower than the one obtained by applying KNN classifier and Fisher classifier methods whether the training set is small or huge. The experiment also demonstrates that SVM can be extremely effective in minimizing the error rate lower than LVQ and BP neural network methods.