World Journal of Engineering Research and Technology

Gift Aruchi Nwatuzie*

ABSTRACT

Intrusion Detection Systems (IDS) play a vital role in defending networks from unauthorized access and malicious ac- tivities. However, traditional IDS methods often suffer from high false positive rates and struggle to identify rare attack types due to significant class imbalance in training datasets. In this study, we implemented an enhanced IDS framework by leveraging machine learning (ML)algorithms Naive Bayes, Support Vector Machine (SVM), and Random Forest while addressing class imbalance using the Synthetic Minority Oversampling Technique (SMOTE). We used[1] dataset, conducted detailed preprocessing and feature selection, and fine-tuned model hyperparameters to improve classification performance. The evaluation demonstrates that the Random Forest model, combined with SMOTE, offers superior results in terms of accuracy and the ability to detect minority classes effectively.

[Full Text Article] [Download Certificate]