论文部分内容阅读
为了提高光滑支持向量机的分类速度和精度,构造一种模糊聚类光滑支持向量机(FCSSVM).运用模糊聚类将训练数据分解为若干子簇,通过引入熵函数近似松弛向量的加函数,并利用最优解处权重向量的表达式导出精确光滑模型;定义测试样本的最近邻子空间,以选择性集成策略组合若干近邻子空间中的分类决策函数.数值实验表明,FCSSVM的分类精度高,迭代次数少,鲁棒性好,分类时间短.
In order to improve the classification speed and accuracy of the smooth support vector machine, a fuzzy clustering smooth support vector machine (FCSSVM) is constructed.Fuzzy clustering is used to decompose the training data into several sub-clusters. By introducing an entropy function to approximate the additive function of the relaxation vector, And the exact smoothing model is derived by using the expression of weight vector of the optimal solution. The nearest neighbor space of the test sample is defined, and the classification decision function in several nearest neighbor subspaces is selectively combined. The numerical experiments show that the classification accuracy of FCSSVM is high , Fewer iterations, good robustness and short classification time.