Bat-Inspired Optimization for Intrusion Detection Using an Ensemble Forecasting Method

An Intrusion detection system (IDS) is extensively used to identify cyber-attacks preferably in real-time and achieve integrity, confidentiality, availability of sensitive information. In this work, we develop a novel IDS using machine learning techniques increase the performance attack process. order cope with high dimensional feature-rich traffic large networks, introduce Bat-Inspired Optimization Correlation-based Feature Selection (BIOCFS) algorithm an ensemble classification approach. The BIOCFS introduced estimate correlation identified features choose ideal subset for training testing phases. Ensemble Classifier (EC) integrate decisions from three different classifiers including Forest by Penalizing Attributes (FPA), Random (RF), C4.5 based on rule average probabilities. integration EC approaches aids handle multi-class unbalanced datasets. proposed evaluated well-known dataset NSL-KDD. experimental results prove that our combined BIOCFS-EC outdoes other relevant methods context appropriate measures. More importantly, decreases time complexity procedure 39.43 2.25 s 16.66 1.28 s, respectively. Also, approach achieves maximum accuracy 0.994, precision 0.993, F-measure 0.992, ratio 0.992 minimum false alarm 0.008% given dataset.