【摘要】 针对一般SVM在机制上难以直接对动态模式进行分类的问题,提出了一种基于函数正交基展开的过程支持向量机.该模型的输入为时变函数,输出为模式类别.在输入函数空间中选择一组适当的正交函数基,将输入函数在该组函数基下进行有限项展开,把展开式系数作为核函数的输入.由于时变函数在基函数映射下与展开式系数一一对应,从而可利用SVM的变换机制实现动态模式分类.给出了基于SMO的求解算法,实验结果验证了模型和算法的有效性.更多还原

【Abstract】 Aiming at the problem that support vector machine(SVM)is difficult to solve dynamic pattern classification directly in mechanism,a process support vector machine(PSVM)model based on orthogonal function basis expansion is presented in this paper.The input of PSVM can be functions with time-varying,and its output can be pattern classifications.A group of proper orthogonal function basis is chosen in input function space.The input functions with finite terms of the function basis are expanded,and the expansion coefficients are considered as the inputs of kernel function.As time-varying functions under basis-function mapping are in one-to-one correspondence with expansion coefficients,the transformation mechanism of SVM is used to implement classification of dynamic patterns.The solving algorithm based on SMO is given,the results of simulation experiments show the effectiveness of the model and algorithm.更多还原