encrypted-traffic

Deep Learning models for network traffic classification

A New version of Python3 botnet, old version: http://github.com/Leeon123/Python3-botnet

Privacy Preserving Collaborative Encrypted Network Traffic Classification (Differential Privacy, Federated Learning, Membership Inference Attack, Encrypted Traffic Classification)

CESNET DataZoo: A toolset for large network traffic datasets

CESNET Models: Neural networks for network traffic classification

PureVPN · vpn software. Practical Windows workflows with a clear interface and fast local response.

This project integrates Explainable AI (XAI) techniques for anomaly detection in encrypted network traffic using ML Algorithms. We employ SHAP (SHapley Additive Explanations) to interpret model decisions and enhance transparency in detecting malicious activities. The system is designed to identify suspicious patterns in encrypted traffic.

Jupyter notebooks with traffic classification examples using CESNET DataZoo and CESNET Models packages

Harness for evaluating encrypted traffic classification under group-aware and temporal splits. Finds that packet size sequences carry 98.5% of the signal, and documents a label-encoding bug that silently degrades results by two thirds.

PCAP → ML tensor extraction for network intrusion detection research.

CATO: Classification of Application Traffic Online – An AI-powered, privacy-preserving network traffic classifier for multi-UE 5G/6G networks. Built with FastAPI, React, and scikit-learn.

Fine-Grained Encrypted Traffic Classification using Graph Neural Networks (GraphSAGE & GAT)

Zero-Shot Malware Traffic Recognition Across Encrypted Protocols using Seq2Vec and Ensemble CNN-GRU

MSc dissertation evaluating targeted website fingerprinting on Tor Browser. Built a traffic collection pipeline and feature extraction framework, then trained Random Forest and SVM classifiers on encrypted Tor traffic to identify both visited websites and browser configuration characteristics, achieving up to 88% accuracy

Iterative leak-surface peeling — a falsifiability-anchored methodology for ML detection of blockchain validator infrastructure attacks. Working draft; v1 ships when arXiv submission lands.

Hyperdimensional Intrusion Detection System for Zero-Day Exploit detection in encrypted traffic using Conformal Geometric Algebra, online learning, and real-time anomaly scoring.

Browser-based harvest-now-decrypt-later (HNDL) threat demo — intercept simulation, post-quantum migration urgency, timeline visualization. Part of crypto-lab.

A Python toolkit for identifying websites from encrypted network traffic using ML on packet metadata (no content decryption).