Utilizing Metaheuristic Ensemble Feature Selection to Enhance Intrusion Detection Systems

Abstract

Intrusion detection plays a crucial role in ensuring the security of computer networks by identifying and preventing unauthorized access or malicious activities. This thesis aim to develop an advanced intrusion detection model by integrating Gravitational search Algorithm (GSA) with Grey wolf Optimization (GWO) algorithm to optimize its performance. The proposed model will combine the strength of GSA-GWO in classification with the optimization capabilities of GWO to enhance the accuracy and efficiency of intrusion detection systems. The research methodology will begin with the selection of appropriate datasets, representative of real-world network traffic, for training and testing the intrusion detection model. Preprocessing techniques will be applied to prepare the datasets, including feature selection and normalization, to ensure the model's robustness and effectiveness. The GSA algorithm will then be implemented and configured, with suitable kernel functions and hyperparameter tuning, to train the intrusion detection model. To optimize the performance of the GSGW-DT model, the Grey Wolf Optimization algorithm will be employed. GWO will mimic the bio behavior of Gey wolves to search for optimal solutions in complex problem spaces. By incorporating GWO, the intrusion detection model will be able to fine-tune BGSA-BGWO parameters and select relevant features, thereby improving accuracy, reducing false positives/negatives, and enhancing overall performance. The developed model will be evaluated using various performance metrics, including accuracy, precision, recall, F1-score, and the area under the receiver operating characteristic curve (AUC-ROC). Comparative analysis will be conducted to assess the superiority of the GSA-GWO model over baseline models or existing intrusion detection approaches. Furthermore, an in-depth analysis of the results will highlight the strengths, weaknesses, and optimization benefits achieved by the proposed model. The conclusion will summarize the key findings and contributions of the study, emphasizing the effectiveness of the GSA-GWO model in optimizing intrusion detection performance. Limitations and potential areas for improvement will be discussed, paving the way for future research. Future work will include exploring additional optimization algorithms, evaluating realtime intrusion detection scenarios, and investigating hybrid approaches to further enhance the model's accuracy and robustness.