基于多类数据分类的改进克隆选择算法

针对传统的克隆选择算法(CSA)只依次单独针对某一类样本数据进行监督学习从而造成分类效率和精确度不高的问题，提出一种基于改进克隆选择算法的多类监督分类算法。算法通过进化学习可以同时获得多类样本数据的最佳聚类中心，进化过程中抗体适度值的计算综合考虑各类的类内相似性和类间差异性，从而保证得到的最佳聚类中心更具代表性。后续的分类实验中，分别利用常用的4组UCI数据和红树林多光谱TM遥感图像对算法进行验证，实验结果表明遥感图像的分类总精度达到92%,Kappa系数为0.91，UCI数据分类结果也较好，证明该算法是一种有效的多类数据分类算法。

Improved clonal selection algorithm for multi-class data classification

The traditional Clonal Selection Algorithm (CSA) can only provide supervised learning for a certain type of sample data, which may result in lower classification efficiency and accuracy, thus a multi-class supervised classification algorithms based on CSA was proposed. This algorithm can obtain optimal clustering center of the multi-class sample data at the same time. The appropriate value of antibodies consider both the same class similarities and the different class differences, thus the best cluster center is more representative. Classification experiments used UCI data and mangrove multispectral TM images, and obtained the overall classification accuracy of 92% and Kappa coefficient of 0.91. UCI data also got a good result. The results show that the algorithm is an effective classification algorithm.