Support Vector Machine And Its Application In Electrocardiogram Classification

This dissertation investigates electrocardiogram classification based on Support Vector Machine algorithm. And this paper puts forward 1-v-1 algorithm of Support Vector Machine for electrocardiogram classification in a creative way. In order to verify the system's stability and creditability, we used American MIT-BIH database to test our algorithms and won higher accuracy. It is better than normal way in constructing algorithm model and classification speed.The purpose, meaning, current situation of ECG classification and research progress of SVM are concerned in foreword.In Chapter One, the general knowledge and measure of electrocardiogram are concerned.Chapter Two is to summarize the basic theory of Statistical Learning Theory and algorithm of Support Vector Machine.In Chapter Three, the problem of multi-classes classification of ECG link with Support Vector Machine is studied. A series of classification methods features and advantages that using Support Vector Machine to classify are put forward.Chapter Four is the core of this paper, in which, the advantage and disadvantage of the series of classification methods are analyzed. We at last choose 1-vs-1 algorithm of Support Vector Machine for electrocardiogram classification. First, model of the algorithm is ascertained. Then the reasonable kernel function and parameter are chosen. We used American MIT-BIH database to test our algorithms and won higher accuracy. After summarizing this paper, a new guideline for the next study is presented.