AI-Driven Cybersecurity: Leveraging Machine Learning For Enhanced Iot Threat Detection And Mitigation

Abstract

Given the vast proliferation of Internet of Things (IoT) devices in our world, these have become an increasingly attractive target to cyber adversaries. It is thus particularly important to continuously be able to assess their overall security posture, detect various anomalous activities, and respond to real-time adversarial attacks leveraging these devices for malicious purposes.

This article presents work in progress on the development of a novel Integrated AI-driven IoT Intrusion Detection Mechanism, called IA2IDM. It works by deploying a Random Forest classifier trained on sets of features generated from an array of datasets from IoT device traffic.Although our preliminary results indicate that achieving a 99% detection rate may not be feasible due to the challenge of having access to adequate training data in the area of IoT networks for defense purposes, we conclude this article by discussing the lessons learned, the promise, and the potential of IA2IDM. It also provides a roadmap on how IA2IDM systems can be developed with sufficient and demonstrable AI confidence and deployed in real-life settings to protect IoT devices now and into the future.