XAI-Driven Hybrid Network Intrusion Detection Agent

The system combines ML-based anomaly detection, expert rule-based inference, and Explainable AI (XAI) to provide transparent and intelligent security decisions. It classifies network traffic as ALLOW, ALERT, or BLOCK based on model probability and handcrafted security rules. Key Features: • Hybrid decision engine combining XGBoost and expert security rules • Explainable AI using SHAP for transparent model predictions • Interactive real-time dashboard built with Streamlit • Probability-based security action system (ALLOW / ALERT / BLOCK) • Professional ML pipeline with preprocessing, feature selection, scaling, and model export • Real-time JSON/manual input for intrusion analysis Tech Stack: Python, XGBoost, SHAP, Streamlit, Scikit-learn, Pandas, Machine Learning, Explainable AI (XAI) This project demonstrates a hybrid AI-driven cybersecurity architecture with interpretable decision-making for real-world intrusion detection systems.