基于支持向量机自适应核的改进算法

支持向量机是在一个向量空间中依靠样本集中的训练向量获取两个样本集之间最优边界的方法.针对支持向量机中基于高斯核的动态自适应核进行了分析,并将传统算法进行改进,通过对点到平面的距离z<,1>的调整,对中间变量a<,1>的细致讨论,以及对迭代停止条件的控制,得到了适合手写体汉字识别系统的新的自适应核算法,并通过其在手写体汉字识别中的实验,验证了新的自适应算法在识别率、泛化误差边界最小化和核参数选择方面的有效性.算法使得在大的核参数空间中没有额外代价的探索变的可行.

Improved support vector machine kernel-adatron algorithm

Support vector machine is a method of obtaining the optimal boundary of two sets in a vector space independently on the probabilistic distributions of training vectors in the sets.The Adatron algorithm in Gaussian radial basis kernel function is analyzed.An improved algorithm is presented for classification in high dimensional spaces.The adjusted z, detailed discussed and limited number of iterations are considered in the new algorithm in order to make adatron-kernel suited for handwritten Chinese character...