Abstract:A support vector machine (SVM) based algorithm for interval number classification models is proposed to cope with the shortcomings of the existing algorithms. Firstly, based on definition of comparison between interval numbers, a classification modeling problem with interval number input is transformed into that with crisp data input and a SVM based interval number linear classification model is derived; Then the linear model with interval number input is extended to nonlinear cases by introducing an appropriate kernel function. The classification models with interval number input have excellent generalization property for small sample sizes and can achieve global solution. An example for quality prediction of continuous casting slabs is analyses and shows the efficiency of the proposed interval number classification algorithm.

A Svm Based Algorithm For Interval Number Classification Models