基于混沌序列的SVM参数选择及其在笔迹鉴别中的应用

基于RBF核的支持向量机(SVM)模型选择取决于两个参数，即惩罚因子和核参数，为了寻找SVM参数的最优组合，利于笔迹鉴别图像的自动识别，提出了基于混沌序列的参数搜索算法以实现SVM模型参数的自动选择。从与网格法和双线性法进行的比较实验可以看出，基于混沌序列的SVM参数选取更简单，更易于实现，并使SVM具有更好的推广能力。在10人笔迹灰度图像库上分类识别实验结果表明，该方法不但可以提高分类识别率，而且显著减少了训练SVM的个数。

Selection of SVM parameters using chaotic series and its application in handwriting verification

In order to find the optimization compound of Support Vector Machine (SVM) parameters, that is penalty factor and nuclear factor, and help to identify the handwriting image, a parameter searching algorithm based on chaotic sequence was proposed to determine the SVM parameters automatically. Compared with the grid search and two-line search, the proposed algorithm is much simpler and easier to be implemented, which makes SVM has better outreach capacity. Classification experiment on 10 people handwriting gray-scale images prove that the proposed algorithm has higher classification rate and significantly reduce the number of training SVM.