Intrusion Detection System based on Fire Hawk Optimizer Algorithm and Deep Learning

Abstract

In light of the increasing cyber threats, intrusion detection systems (IDS) have become critical for protecting computer systems. This thesis introduces an advanced IDS that combines the Fire Hawk Optimizer (FHO) algorithm with a one-dimensional Convolutional Neural Network (CNN-1D) to improve accuracy and efficiency in detecting intrusions. The FHO algorithm is used for feature selection, reducing input data dimensionality while retaining essential features. These optimized features are then analyzed by the CNN-1D model, which identifies temporal and spatial patterns in network traffic. The system's performance is evaluated using the CSE-CIC-IDS2018 dataset, demonstrating high accuracy, recall, precision, and F1 scores in detecting various network attacks. The FHO algorithm enhances efficiency by reducing computational overhead and training time without sacrificing detection performance. Future research will involve testing the system on additional datasets and exploring its deployment in real-world network environments.