A new clonal selection algorithm based on radius regularization of anomaly detectors

Nowadays anomaly detection based on Artificial Immune system (AIS) is highly regarded among the research community and Clonal selection algorithm is one of the most interesting fields of it. In this paper we model self-region by a number of spheres with fix radius. We have assumed anomaly detectors as spherical shape with random central point and variable radius out of this region. Detection rate (DR) and false alarm (FA) are influenced by radius of Detectors. Large radiuses would increase both DR and FA and small radiuses would decrease them, so selecting appropriate quantity plays an important role for achieving high accuracy in recognition. The radius of detectors is determined based on self-environment. The radius of self-region spheres has straight effect on detectors radiuses and is an important parameter to achieve high accuracy. As a result of finding imperfect self-region radius we have faced weak detectors that couldn't recognize anomaly perfectly. In this paper we propose a novel algorithm to achieve suitable accuracy independent of self-region radius using learning automata.