基于竞争学习型否定选择算法的孔洞检测

Detecting Holes by Competitive Learning Negative Selection Algorithm

为了检测人工免疫系统中存在的孔洞,提出一种基于竞争学习的否定选择算法。该算法依据自体集和非自体集产生两个检测器集合,分别对非自体和孔洞进行检测。借鉴竞争型神经网络的特点,检测器之间构成一种竞争学习的关系。针对工程应用中自体集和非自体集可能动态变化的情况,检测器集合会根据当前自体集和非自体集自动更新。仿真实验表明,该算法可以有效地检测出孔洞。

A negative selection algorithm based on competitive learning was proposed in order to detect holes in artificial immune system. Holes and non-selves were detected by two detectors sets that were generated according to selves and non-selves. Drawing lessons from competitive neural networks, the two detectors sets will be updated automatically in the light of selves and non-selves. The results of experiments prove that the algorithm can detect holes effectively.