| 多类支撑向量机在中医舌质、舌苔分类和识别的应用研究 |
| |
| 引用本文： | 张新峰,沈兰荪,卫保国,蔡轶珩.多类支撑向量机在中医舌质、舌苔分类和识别的应用研究[J].电路与系统学报,2004,9(5):110-113. |
| |
| 作者姓名： | 张新峰 沈兰荪 卫保国 蔡轶珩 |
| |
| 作者单位： | 北京工业大学,信号与信息处理研究室,北京,100022 |
| |
| 基金项目： | 国家自然科学基金资助项目(60301003，60227101) |
| |
| 摘 要： | 在分解法进行多类支撑向量机的训练中，所有的两类问题训练都采用一种核函数是不合适的。本文在一对一(OAO，one against one)训练算法中优先选择线性核函数，并将训练结果用有向无回路图DAG(Directed Acyclic Graph)的方法进行中医舌质、舌苔识别。结果表明，该方法可以获得超过85％的正确识别率，识别速度也比较快。
|
| 关 键 词： | 核函数 多类支撑向量机 一对一 有向无回路图 |
| 文章编号： | 1007-0249(2004)05-0110-04 |
| 修稿时间： | 2004年3月11日 |
Application of the Multi-class SVM to the Classification and the Recognition of Tongue Substance and Tongue Coat |
| |
| ZHANG Xin-feng,SHEN Lan-sun,WEI Bao-guo,CAI Yi-heng.Application of the Multi-class SVM to the Classification and the Recognition of Tongue Substance and Tongue Coat[J].Journal of Circuits and Systems,2004,9(5):110-113. |
| |
| Authors: | ZHANG Xin-feng SHEN Lan-sun WEI Bao-guo CAI Yi-heng |
| |
| Abstract: | It is not appropriate for all the bi-class questions to select the same kernel function during the training of multi-class support vector machine by the decomposition method. In this paper, the method of selecting the linear kernel function is primarily used to the training algorithm of OAO(one against one)and the trained result thus attained is applied to the recognition of tongue substance and tongue coat in the traditional Chinese medical diagnosis by DAG(Directed Acyclic Graph). Experience shows that the recognition rate by this method exceeds 85 percent with increased speed of recognition. |
| |
| Keywords: | kernel function multi-class SVM one against one directed acyclic graph |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
