一种基于虚拟力导向和细胞分化的免疫网络分类算法

针对传统免疫网络分类算法中抗体细胞进化缺乏有效指导的问题,提出了一种基于虚拟力导向和细胞分化的免疫网络分类算法(VCAINC).算法引入虚拟力场区的概念,基于虚拟力场区定义抗体所受的虚拟作用力以指导抗体的进化过程,并根据移动收敛条件判定抗体的合理位置;对于无法达到收敛条件的抗体,采用细胞分化策略提高免疫网络的分类性能;分析了算法在UCI标准数据集的分类效果.实验结果表明,VCAINC对于多个标准数据集均具有良好的分类性能,能够有效指导抗体细胞的进化.

An Artificial Immune Network Classification Algorithm Based onVirtual Force Direction and Cell Differentiation

Abstract： Aimed at the lack of effective methods to guide the evolution of antibodies in the traditional immune network classification algorithm, a virtual force and cell differentiation based artificial immune network classification algorithm (VCAINC) was proposed.The algorithm introduced the definition of virtual force zone to guide the evolution of antibodies.Then, the moving convergence condition was introduced to determine the most suitable position of antibodies.For the antibodies which cannot satisfy the moving convergence condition, the strategy of cell differentiation was utilized to optimize the classifier. Finally, the performance of classification on UCI dataset was analyzed.The experimental results show that the VCAINC can effectively guide the evolution of antibodies and improve classification performance significantly.