Comparison And Application Of Support Vector Machine And Orthognal Design Method

The statistic learn theory and the method of supporting Vector mechanismpattern identificatton base on it are described aiming to the hypothesis of big numbertheorem in traditional pantern identifaction theory.Presents the design methods ofclassification implement according to the minimum structure risk,namely accordingto both optimization of classification ability and minimum of experience risk. Thesupporting vector mechanism is the concrete realization of VC dimension theory ofstatistic 1earning theory and minimum structure risk principle which transform theincome space into a high dimension space through nonlinearity transformation andseek the optimum linearity classification facet.The purpose of this paper is to provide a discussion and comparison of SupportVector Machine (SVM) and the orthogonal design method. SVM is a populartechnique for classification and regression. Orthogonal test design is a scientific andeffective method in planning experiment, and it is extensively applied in chemistryindustry. This paper, using 2 factors and 7 dimensions orthogonal test (dryingexperiment) as an example, shows how to use the support vector machine toorthogonal test table and support vector machine′s applications on test design. Theorthogonal test table was used to studying the optimal conditions of experiment, andthe significant factors. The paper also gives the result of applying support vectormachine to the testing result of orthogonal test design. By the comparison, it has beenproved that the prediction ability of support vector machine is more powerful thanorthogonal design method in the drying experiment. Support vector machine has apromising application in experimental design.