非线性回归支持向量机的SMO算法改进

为了解决非线性数据和非线性函数的回归问题，采用了支持向量机序列最小优化算法.原始序列最小优化（SMO，Sequential Minimal Optimization）算法存在训练速度慢和训练结果不稳定的缺点，为了能加快SMO算法的训练速度和提高训练结果稳定性，通过改进优化乘子更新方法、采用双阈值法、预存核函数、增加停机准则等方法对SMO算法做了改进.仿真实验表明，改进的算法能很好地对非线性数据和非线性函数进行回归，具有比原始SMO算法更快的训练速度和稳定的训练结果.

Improved SMO algorithm of nonlinear regression support vector machine

In order to solve the regression problems of nonlinear data and nonlinear function, the support vector machine (SVM) sequential minimal optimization (SMO) algorithm was adopted. The original SMO algorithm has deficiencies such as low training speed and instability training results. To accelerate the training process of SMO algorithm and promote training stability of the solution, the SMO algorithm was improved by updating the optimization multipliers method, using double threshold values, caching kernel function outputs, adding stop criterion. Simulation results show that the improved algorithm performs well for regression of nonlinear data and nonlinear function, and it has faster training speed and better training result stability than original SMO algorithm.