Biological Feature Selection and Classification Techniques for Intrusion Detection on BAT

Privacy is a significant problem in communications networks. As a factor, trustworthy knowledge sharing in computer networks is essential. Intrusion Detection Systems consist of security tools frequently used in communication networks to monitor, detect, and effectively respond to abnormal network activity. We integrate current technologies in this paper to develop an anomaly-based Intrusion Detection System. Machine Learning methods have progressively featured to enhance intelligent Anomaly Detection Systems capable of identifying new attacks. Thus, this evidence demonstrates a novel approach for intrusion detection introduced by training an artificial neural network with an optimized Bat algorithm. An essential task of an Intrusion Detection System is to maintain the highest quality and eliminate irrelevant characteristics from the attack. The recommended BAT algorithm is used to select the 41 best features to address this problem. Machine Learning based SVM classifier is used for identifying the False Detection Rate. The design is being verified using the KDD99 dataset benchmark. Our solution optimizes the standard SVM classifier. We attain optimal measures for abnormal behavior, including 97.2 %, the attack detection rate is 97.40 %, and a false-positive rate of 0.029 %.