Time Series Data Pattern Classification using Fuzzy Membership Functions and Support Vector Machines - KOSPI 200: Korea Composite Stock Price Index 200

摘要

SVM (Support Vector Machine) is a binary classifier proposed by Vapnik. SVM has proven performance in various application fields by minimizing misclassification based on mathematical theories. Recently, FSVM that applies Fuzzy membership function to SVM has been proposed. In this study, it is proven that FSVM (polynomial kernel) has reduced learning time better than SVM when fuzzy membership functions of FSVM have been expanded from 2-dimension to bigger than 3 dimension.