Intrusion Detection System (IDS) Project 🚀 Overview This project focuses on building a robust Intrusion Detection System (IDS) that detects and classifies network intrusions. Two machine learning models are developed:

Attack Classification Model: Multiclass classification for 12 types of network attacks. Attack Detection Model: Binary classification (attack vs. no attack). Both models employ advanced feature selection techniques, preprocessing pipelines, and ML/DL algorithms to achieve high accuracy and performance.

🎯 Objectives Attack Classification Model:

Multiclass classification of 12 attack types. Dataset: CIC-DS with 2.3 million records and 38 preprocessed features. Attack Detection Model:

Binary classification for attack vs. no attack. Dataset: UNSW-NB15 with 1.5 million records and 50 raw features (preprocessed). 🌟 Features Ensemble Feature Selection: Combines multiple feature selection methods (Chi-square, RFE, etc.) via majority voting to identify the most relevant features. Comprehensive Preprocessing: Handling class imbalance (SMOTE/undersampling). Scaling and normalization of features. Dimensionality reduction (Ensemble_ feature_selection). Model Implementation: Algorithms: MLP, Random Forest, XGBoost, Decision Tree. Comparative evaluation of models. Performance Visualization: ROC curves. Confusion matrices. Metrics comparison (accuracy, precision, recall, F1-score, ROC-AUC).