论文部分内容阅读
支持向量机是最近几年发展起来的一种新的智能学习方法,以RBF为核函数的支持向量机在实际应用中表现出良好的学习性能,被广泛的应用到模式识别中,其参数C和σ对SVM的性能起决定性作用。文中阐述了SVM原理,给出了RBF-SVM的性能随参数变化的规律,并得到了参数C和σ对SVM支持向量个数和测试样本错分率的影响曲线图。最后,通过采用不同核函数的SVM对电力通信网光纤保护通道进行风险评估,比较其评估性能,同时验证了RBF-SVM性能的优势。
Support vector machine (SVM) is a new intelligent learning method developed in recent years. The RBF-based SVM shows good learning performance in practical applications and is widely used in pattern recognition. The parameters C And σ have a decisive effect on the performance of SVM. In this paper, the principle of SVM is expounded, the performance of RBF-SVM is changed with the parameters, and the influence of parameters C and σ on the number of SVM support vectors and the misclassification rate of test samples is obtained. Finally, the risk assessment of optical fiber protection channel of power communication network is carried out by using SVM with different kernel functions to evaluate the performance of the optical fiber protection channel. At the same time, the advantages of RBF-SVM performance are verified.