Adaboost 와 SVM을 결합한 앙상블 classify를 이용한 알치하이머 질병군 과 일반군 인지장애 환자군의 분류

Abstract

Functional magnetic resonance imageing(fMRI), that measure blood oxygen level- dependent signal, have been suggested as useful tool for distingushing diverse diseases associateed with brain function. disease associate with function of brain is Alzheimer disease(AD) and Mild cognitive impairment(MCI). we will introduct classification method in paper. classification in functional magnetic imaging(fMRI) is challenging because fMRI data is high dimension, small number of available data set. general method of classification was used single method but we will use ensemble classification method to improve accuracy and the method has high operation speed for fMRI data analysis. The method is combined form Support Vector Machine(SVM) and Adapt Boost(Adaboost) method. First the method is support vector machine(SVM), it was known as great the ways in compute vision field and as guaranteed high accuracy. Second Adapt Boost method is for impoving accuracy of any given learning algorithm. The Adaboost method was consisted of weak classifier and strong classifier, we was used support vector machine as weak classifier, single support vector machine(SVM) output was applied by a weighted aggregation of next classify's output. An Adaboost technique was applied, modified to look for optimal weighted aggregation of classifier. In study, we was used data set, which is normal control(NC), Alzheimer disease(AD) and Mild cognitive impairment(MCI) data set and to obtain correlation coefficient in brain network, we was used functional region of interation(ROI), which is Basal Ganglia network, Default Mode network, Saliencenetwork, Motor, that is feature for classification. we had operated classification method with the features. feature for classification is very impotant to result so we will congitantingly decide it.