一种改进的RNS算法的研究及应用

Study and application of an improved RNS algorithm

针对异常检测问题,改进了混合神经元免疫系统中实数阴性选择算法(RNS,real-valuednegative selection),使其更好地应用在多种纬度的空间上。实验结果表明,使用改进后的实数阴性选择算法(GRNS,general RNS)的神经元免疫系统仅用很少的自体集,就可以达到较高的检测率和比较低的误警率,在高维度特征量的异常检测中取得了很好的效果。

For anomalous detection,we improved the real-valued negative selection（RNS） algorithm of hybrid neuro-immune system（HNIS）,and named it general RNS（GRNS）.The algorithm is applied to neurons-immune system for anomalous detection.The results show that,with such small number of self samples,the improved NIS has high detection rates and low false alarm rates.