A novel optimized probabilistic neural network approach for intrusion detection and categorization
Nadir Omer, Ahmed H. Samak, Ahmed I. Taloba, Rasha M. Abd El-Aziz
Abstract
Nowadays, the web provides all of the nation's daily necessities, and time spent online is rising quickly. The Internet is being used more widely than ever. As a result, cyberattacks and cybercrime are becoming more prevalent. Various machine learning techniques will be used to recognize network attacks and defend against cyber security threats. Developing intrusion detection systems can improve cybersecurity and identify anomalies on a computer server. An efficient intrusion detection and prevention system will be created using machine learning techniques. Each intrusion detection categorization system evaluated in this study has its unique uses. The Firefly Optimization (FFO) technique was used to detect the intrusions before the categorization procedure was carried out using a machine learning classifier. It considered how the anomalies in networks were categorized in this research. The outcomes of the detection techniques will be validated using the Knowledge Discovery Dataset (KDD-CUP 99). The proposed method involves Probabilistic Neural Network for the categorization. The implementation will assess many performance metrics for various cyber-attack types, including specificity, recall, F1-score, accuracy, precision, and sensitivity. The proposed technique achieves a high accuracy of 98.99%.