Development of an AI-Powered Intrusion Detection System Using Logistic Regression
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Malaysia Sarawak
Abstract
Description
The fast rise in cyberattacks has highlighted the vital need for effective and accurate Intrusion Detection Systems (IDS) to secure modern network infrastructures. Traditional intrusion detection systems, including signature-based or anomaly-based systems, frequently have large false positive rates, limited adaptability, and trouble identifying complex attacks. Using Logistic Regression, a machine learning method noted for its ease of use and efficiency in binary classification problems, this project suggests creating an AI-powered intrusion detection system. Data preprocessing for the project includes handling missing values, feature scaling, and class imbalance using SMOTE. A publicly accessible network traffic dataset (such as NSL-KDD) will be used to train and evaluate the logistic regression model. The model's effectiveness in identifying and categorising intrusions will be evaluated through performance evaluation utilising metrics including accuracy, precision, recall, F1 score, and confusion matrix. The expected outcome is a reliable and scalable IDS capable of improving detection rates while minimizing false positives. This project aims to contribute to the field of cybersecurity by improving network security and offering an affordable solution that can be tailored to different network environments by utilising machine learning.