Proposal Resubmission – 15th April 2025

Today, I received full approval for my final year project proposal titled “Comparative Analysis of Unsupervised Machine Learning Models for Network Intrusion Detection Systems” after addressing minor revisions suggested by the review evaluators.

Evaluators' Comments:

Research Objectives:

• Previous Version: The research objectives were not structured according to research methodology.

• Update: The research objectives now includes methodology-based objectives as follows:

  • To review existing unsupervised ML-based NIDS technologies and optimize unsupervised ML algorithms to perform unknown anomaly detections in an NIDS simulation 

  • To design an unsupervised ML-based NIDS framework to perform an unknown anomaly detection experiment

  • To develop an unsupervised ML-based NIDS simulation using Isolation Forest (IF), Local Outlier Factor (LOF), and K-Means Clustering algorithms using CSE-CIC-IDS2018 and UNSW-NB15 datasets 

  • To assess and compare each unsupervised ML algorithm’s performance in unknown threat detection using 10-Fold Cross Validation technique, precision, recall, F1-score, and AUC-ROC evaluation metrics, and Wilcoxon Signed-Rank test

Methodology:

• Previous Version: The methodology was generally outlined but lacked clarity in terms of the process flow and specific NIDS framework to be used.

• Update: The methodology was reorganized for better clarity:

  • Data Preparation: Included a detailed explanation of preprocessing steps (feature selection, handling missing values, normalization).

  • Model Selection: Clear justification for choosing IF, LOF, and K-Means, based on their unsupervised learning abilities to detect anomalies in unlabelled data.

  • Evaluation Metrics: Expanded on metrics (precision, recall, F1-score, AUC-ROC) and the reasoning behind their selection.

  • Statistical Tests: Added justification for using the Wilcoxon Signed-Rank test to compare model performance with statistical significance.

Achievements:

• Addressed evaluators' comments and successfully obtained full approval.

• Refined the proposal to meet the academic and formatting standards.

• Ensured the project aligns with both academic expectations and real-world applications.

Upcoming Plans:

• Begin the literature review and theoretical framework for Chapters 1 and 2.

• Draft content for Progress Report 1 in preparation for supervisor review.