Ecg Classification And Neural Networks, Vc Dimension

As a routine medical check-up, ECG is of importantly clinical value; especially for heart chirurgery; but the traditional method of false diagnosis is limited by individual professional knowledge and clinical experience; at the same time the speed of false diagnosis is very slow. However now the computer has been applied broadly in signal processing; so it is meaningful to establish a new method based on the computer to classify ECG fast and exactly .In this paper support vector machine (SVM) and probabilistic neural network (PNN) are used to establish a new method to classify ECG. The main work of the paper:(1) With the multi-resolution analysis of wavelet, the different frequencyingredients are distinguished; then the main characters of ECG is built up. Afterward the ECG classification is implemented with SVM.(2) A new BP neural network is put forward to be used to realize theclassification of ECG; especially, a neural network is constructed to be used to realize the muliclassification of ECG.(3) From a geometry angle, the Vapnik-Chervonenkis dimension of feed forward neural network is discussed.; especially , the VC dimension of nonlinear neural network is discussed with the hypersurface arrangement.New creations of the paper:Constructing new BP neural networks to classify the ECG, especially a neural network be used for multi-classification of ECG is of a certain practical value.Discussing the VC dimension of neural network with the hypersurface arrangement.