Generating an Approximately Optimal Detector Set by Evolving Random Seeds

The detector generation algorithm is the core of a negative selection algorithm (NSA). In most previous work, the NSAs generate the detector set randomly, which cannot guarantee to obtain an efficient detector set. To generate an approximately optimal detector set, in this paper, a novel detector generation algorithm for the real-valued negative selection algorithm (RNSA) is proposed. The proposed algorithm, named as the EvoSeedRNSA, adopts a genetic algorithm to evolve the random seeds to obtain an optimized detector set. The experimental results demonstrate that the EvoSeedRNSA has a better performance.