基于支持向量机的步态分类方法

针对小样本步态数据引起的分类器泛化能力差的问题,提出了基于支持向量机的步态分类方法.采集了24名青年和24名老年受试者的步态数据,提取24个步态特征训练支持向量机,采用交叉验证方法评估分类器的泛化性能.结果表明,本文提出的方法能够有效地对小样本步态数据分类,并且具有良好的泛化性.不同的核函数对分类性能影响较小.与传统反向传播学习算法的神经网络分类器进行了比较,支持向量机分类性能明显优于传统反向传播学习算法的神经网络.支持向量机在步态分类中具有广泛的应用前景.

Approach for Gait Classification Based on Support Vector Machines

To solve the problems of small sample size and poor generalization ability in the classification of gait data,this paper investigated the application of support vector machines(SVMs) to the gait classification.The gaits of 24 young and 24 elderly participants were recorded during their normal walking.Altogether,24 gait features describing the gait patterns were extracted for developing gait classification models and evaluating the classification generalization performance by cross-validation.The results indicate that the proposed approach can classify small sample size gait data effectively with good generalization performance on which the different kernel functions has a slight impact.In addition,the improved classification performance of the SVM is evident compared with the neural network based on standard Backpropagation(BP) learning algorithms.These results demonstrate considerable potential in applying SVMs to gait classification.