基于凸壳的支持向量机训练集选取算法研究

基于对支持向量机训练算法的研究与理解,针对支持向量的几何特点,结合凸壳的几何性质,对支持向量机训练算法的工作集选取进行了改进.在LIBSVM的环境下进行了对比分析.数值实验表明,改进的基于凸壳的训练数据选取算法缩短了训练时间并保持了良好的泛化能力,具有一定的参考价值.

Research on Algorithm for Training Set Selection of Support Vector Machines Based on Convex Hulls

For support vector geometric characteristics, the algorithm for training sets selection of support vector machines is improved combination of the geometric properties of Convex Hull based on support vector machine training algorithm to study and understanding. And it carried out a comparative analysis in Libsvm. Numerical experiments show that the improved Convex Hull-based training data selection algorithm reduces the training time, maintains a good generalization ability and has some reference value.