Kernel Function Studies on the Support Vector Machine in Lower Limb Motion Pattern Recognition of Stoke Patients

Learning algorithms of the support vector machine is to map the input vector to a high dimensional space through certain kernel function and separate the image of the original linear input vector with the maximum of interval under consideration. This paper is about the limb motion recognition problem of stroke patients, mapping the input vector to the reproducing kernel RKHS (reproducing Kernel Hilbert space) space and using the methods in linear space to solve nonlinear problems. Meanwhile, feature transformation is achieved by defining the inner product of samples in the feature space after its characteristics are changed. Experimental results show that the support vector machine which is made up of new Kernel function can greatly improve the recognition rate of action under the conditions of Mercer, providing theoretical basis for modeling of lower limb rehabilitation training system of stroke patients.