在提升支持向量机分类算法精度的问题上,提出了一种基于混合高斯模型和空间模糊度的支持向量机算法。该算法通过采用多维混合高斯模型的求带分布密度概率函数的二次规划问题的最优解,减少不同的输入样本数据对分类超平面造成的影响,引入了优化后的空间模糊度因子和空间模糊度函数。在实验仿真上,采用了人工选择的样本数据集和UCI机器学习数据库中的样本数据集进行多次实验,最后通过对比分析的方法突出了算法在分类精度上的优势。
For the problem that to enhance the accuracy of SVM classification algorithm,this paper proposed an optimal solution based on Gaussian mixture model and spatial ambiguity SVM algorithm. This algorithm used multi-dimensional Gaussian mixture model with quadratic programming problem to seek the probability density function,reduced the different input sample data to hyperplane and introduced the optimized space fuzzy factor and spatial ambiguity function. In the simulation experiments,it used artificial selected sample data sets and UCI machine learning database sample data sets. It finally by the comparative analysis,it highlights the advantages of the proposed method on the classification accuracy.