Genetic algorithm with different feature selection techniques for anomaly detectors generation

摘要: Intrusion detection systems have been around for quite some time, to protect from inside ad outside threats. Researchers and scientists are concerned on how enhance the intrusion performance, be able deal with real-time attacks detect them fast quick response. One way improve performance is use minimal number of features define a model in that it can used accurately discriminate normal anomalous behaviour. Many feature selection techniques out there reduce sets or extract new them. In this paper, we propose an anomaly detectors generation approach using genetic algorithm conjunction several techniques, including principle components analysis, sequential floating, correlation-based selection. A Genetic was applied deterministic crowding niching technique, generate set single run. The results show sequential-floating best results, compared others tested, especially floating forward accuracy 92.86% train 85.38% test set.