Application Of Convex Hulls In Support Vector Classification

Statistical learning theory (SLT) is based on the principle of the structural risk minimization (SRM), and it is a new set of theory, which especially aims at machine learning issues under the circumstances of rare samples. Based on the SLT, supporting vector machine (SVM) method has been developed as a new machine learning algorithm and is a specific use in SLT. At present, it is mainly used in the fields of pattern recognition, regression and probability density function estimation, etc.Some important concepts about SLT are firstly introduced . In addition, the concept and process of solution of SVM is also introduced to illustrate that SVM is a sort of convex optimization issue whose solution is characterized by global optimum. Through the analysis of SVM principle, convex hull theory is used for supporting vector machines, with acme-set of convex hull replacing the entire sample set for training, and then IRIS data set is used to do the simulation. The simulation results show that this method has the same performance as learning by the entire sample set, but reduces the storage space and improves the learning speed.At the end of the essay, the application of SVM based on convex hulls in classification of fault diagnosis system of turbo generator unit is discussed. After the pretreatment and feature selection and extraction in the process of classification are described in detail, experiments with actual data show the validity of the algorithm.