Unsupervised Anomaly Detection for Unseen Cyberattacks in IoT Networks

Abstract

This project evaluates unsupervised machine learning methods for detecting previously unseen cyberattacks in IoT network traffic. Using the CICIoT2023 dataset, Isolation Forest, Local Outlier Factor, and One-Class SVM are compared under proxy zero-day scenarios, focusing on detection performance, false-positive rates, and computational efficiency.