一种改进的树突状细胞算法及其性能稳定性实验

摘要

Aiming at the problems of original Dendritic Cell Algorithm (DCA)that it has misclassifications when the context changes,and its error rate is proportional to the times of state changes,we improve it based on the original algorithm by the strategies such as setting dynamic threshold,adding the evaluation factor of subsequent antigen on current sampling antigen and the inflammation signal with the function of amplifying the effects of other signals,etc.Then we apply the modified algorithm to the standard dataset Breast Cancer. Experimental result shows that compared with original DCA,this algorithm has amelioration in both stability and recognition rate.%针对原有树突状细胞算法DCA（Dendritic Cell Algorithm）在环境状态转变时存在误分类，且随着状态转变的次数越多错误率越高的问题，在原算法的基础上通过设置动态阈值，加入后续抗原对当前采样抗原的评价因子，以及具有放大其他信号功能的发炎信号等策略对其进行改进，将改进算法应用于标准数据集Breast Cancer。实验结果表明，与原DCA算法相比，该算法的稳定性和识别率都有一定改善。