rule) generated by CGREA only contains two or three attributes, and the above principles are almost satisfied before pruning.
In our previous work , the number of optimal detectors generated by CGREA, on which AMCS is based, is small and simple.
Testing on the 10% Training Set, the coverage of a single detector in the optimal detector sets, which is generated by AMCS and CGREA respectively, are compared in Fig.
Because CGREA evaluates detectors considering the coverage probabilities of attacks and the approximate degree of distributions of the training set and the covered set, it can be seen from Fig.
Furthermore, the proposed avidity function, which is capable to balance the weight of NS and positive selection in AMCS via balance factor, is more applicable to IIDS than that of CGREA.