基于人工免疫网络记忆的新型分类器研究

该文首先简要介绍了自然免疫系统的免疫记忆原理,之后对aiNet和AIRS机制进行了分析,指出免疫记忆在两种算法中起关键作用。aiNet利用传统聚类技术对产生的记忆抗体进行数据聚类分析,无法得知原始抗原类别。AIRS通过训练产生记忆细胞池,利用最近邻原理对原始抗原分类。文中,aiNet的记忆抗体生成机制与KNN分类机制结合,提出基于人工免疫网络记忆的新型分类器AINMC———人工免疫网络记忆分类器。实验结果表明,所提出的新型分类器具有良好的记忆和泛化性能,准确率可以与许多传统方法相比较。

The Research of New Classifier Based on Artificial Immune Network Memory

In the paper,the principle of immune memory is briefly introduced at first.Then the mechanisms of aiNet and AIRS are analyzed.It is pointed out that immune memory is the key of the two algorithms.The traditional technologies of cluster are used in aiNet to analyze the cluster of memory antibodies.It cannot get the classes of original antigens.AIRS results in the memory cell pool after it is trained and classifies the original antigens by KNN.The mechanism of producing memory antibodies and KNN are combined in this paper.A new classifier based on artificial immune network memory(AINMC) is presented.The results of test show that the presented classifier has good performance on memory and generalization.The accuracy can be compared with many traditional methods.