Study on extension negative selection algorithm

Aimed at the problems of low generation efficiency, serious redundancy and poor matching capability of detectors, the extension negative selection algorithm (ENSA) is proposed by fusing extenics and artificial intelligence system. The basic conceptions of ENSA are described by basic element, and the affinity between detector and antigen or antibody is calculated by dependent function. The algorithms of extension detector generation and optimisation are designed, and the parameters of them are analysed. Furthermore, the performance of ENSA is analysed both in theory and simulation experiment. The results from the Iris dataset show that when generating five detectors, the coverage rate of ENSA is 87.5% which is 70.28% higher than that of RNSA and 76.95% higher than that of V-Detector algorithm; when the expected coverage rate is 90%, three detectors are required in ENSA, which is 14 fewer than that of RNSA and 74 fewer than that of V-Detector algorithm; when the same antibodies are tested, the correct rate of ESNA and RNSA is 100% while the VDetector algorithm’s is 90%.