Hybrid Feature Selection for Effective Intrusion Detection

Network Security is a major challenge due to the rapid growth and expansion of modern networks. There is an imminent need to protect organizations from malicious attacks by using intrusion detection systems. Conventional IDS techniques may not be able to identify new intrusions, hence building an IDS with existing intrusion detection strategy becomes essential. In this paper a hybrid feature selection for a robust IDS is proposed. It finds optimal features from an effectively preprocessed data to detect attacks successfully. Our proposed model is validated using benchmark dataset UNSW_NB15 and its performance is compared to various existing approaches and found that our model outperformed them.