Comparative Analysis to Develop a Dimensionality Reduction Model for Classifying Intrusion Detection Systems

This study aims to compare the performances of different combinations of dimensionality reduction techniques and classification algorithms for intrusion detection based on forensic frameworks. The study used a real-world dataset collected from a forensic framework to evaluate the performances of four different combinations of dimensionality reduction techniques and classification algorithms namely; Improved Genetic Algorithm (I-GA) and Support Vector Machine (SVM), I-GA with K-Nearest Neighbor (KNN), Locally Linear Embedding (LLE) with SVM, and LLE with KNN. The dataset was preprocessed and normalized then divided into training and testing sets to conduct performance evaluations based on accuracy, precision, recall, and F1 score. The results of the study showed that the combination of I-GA with SVM had accuracy of 95.17%. The combination of LLE with SVM had an accuracy of 65.88%, while the combination of I-GA with KNN had an accuracy of 98.73%. The combination of LLE with KNN had 95.12%. The study's findings suggest that the combination of I-GA with KNN is the most effective combination for detecting potential intrusions in network data. The results also demonstrated the importance of using dimensionality reduction techniques in intrusion detection systems to increase efficiency and accuracy. The study's findings could contribute to the development of more efficient and accurate intrusion detection systems to enhance the security of computer networks.