Cyber Security Detection System Using Machine Learning

Abstract: This project focuses on developing a Cyber Security Detection System that utilizes various machine learning models to classify network traffic as either normal or malicious. The system preprocesses network traffic data, performs feature analysis, and trains models to detect different types of attacks. Key features include dataset handling, where network traffic data is read and pre processed, followed by feature engineering that examines categorical variables such as protocol type, login success, and attack distribution. The project implements several machine learning models, including Gaussian Naive Bayes, Decision Tree, Random Forest, Support Vector Machine (SVM), Logistic Regression, Gradient Boosting Classifier, and Artificial Neural Networks (ANN). Performance analysis of the models reveals high accuracy, with the best model achieving a training accuracy of 99.88% and a testing accuracy of 99.88%. The classification report shows excellent precision, recall, and F1-scores for various attack types, including Denial of Service (DoS) and normal traffic, both achieving 100%. Although detection rates for U2R attacks are lower due to fewer samples, the system demonstrates significant overall effectiveness in identifying other attack types such as Probe, R2L, and DoS attacks. Additionally, the system includes user management features, such as user registration with OTP verification, admin approval for login, and admin notifications for detected attacks. The system also offers user profile management, real-time attack detection input, and a feedback system to improve overall performance. Admins can analyze feedback to further enhance the system.

Keywords: Cyber Security, Machine Learning, Network Traffic, Intrusion Detection, Feature Engineering