cicids-2017

Journal Article: International Journal of Information Technology

Hybrid NIDS combining unsupervised autoencoders (Feedforward, Denoising, Convolutional) with a PPO reinforcement learning agent trained on 2.8M+ CICIDS-2017 flows. Achieves ~92% F1-score with cost-sensitive reward shaping prioritizing attack recall.

This research proposes a hybrid approach using Principal Component Analysis (PCA) for dimensionality reduction and Support Vector Machines (SVM) for classification. The goal is to optimize the trade-off between detection accuracy and computational efficiency, making real-time detection feasible on resource-constrained environments.

AI-powered network intrusion detection using Random Forest + Isolation Forest on CICIDS-2017

This project implements a machine learning–based Intrusion Detection System (IDS) designed to identify and classify malicious network traffic. Leveraging the CICIDS2017 dataset, which contains realistic and diverse attack scenarios, the system applies supervised learning techniques to detect anomalies and cyber threats.

Intrusion Detection with ML: CICIDS-2017 → Preprocessing → XGBoost → PCA → Real-Time Power BI Dashboards