Classification of Anomaly Detection Attacks in IoT Devices using Machine Learning

Abstract

Concerns about the security of Internet of
Things (IoT) devices are growing, and the abstract
emphasizes how vulnerable they are to anomaly
attacks. The project suggests a way to find strange
things in the Internet of Things (IoT) using Machine
Learning (ML) methods like Support Vector Machine
(SVM) and Random Forest (RF), along with votes
and stacked models. Experiments using the NSL-
KDD dataset show that RF and stacked models can
get high accuracy rates with few false positives. RF
significantly beats current material, which shows how
useful it could be. Some ensemble methods, like
Voting Classifier (RF + AB) and Stacking Classifier
(RF + MLP with LightGBM), are very good at
detecting and preventing anomalies because they
have high accuracy, memory, and precision. In
addition, the project includes user testing through a
front end built on the Flask framework and user
identification, which makes IoT anomaly detection
more useful in real life.