Binary and Multi-Class Classification on the IoT-23 Dataset
Alia Ahli, Ayesha Raza, Kevser Ovaz Akpinar, Mustafa Akpınar
Abstract
In the recent years, Internet of Things (IoT) technology has become widespread and is commonly used by organizations for many different purposes. With the increase in the demand for this technology, new security challenges arise. It becomes difficult for cybersecurity experts to administer security measures and prevent the IoT network from malicious attacks. This paper proposes the use of machine learning algorithms – Random Forest (RF), Multi- layer Perceptron (MLP) and Gradient Boosting (GB) to detect malicious traffic flows and anomalies in the IoT network. For this purpose, Aposemat IoT-23 dataset is used, and the models are built, trained and tested on it. Two types of classification are considered – binary and multi-label. Random Forest and Gradient Boosting classifier were able to achieve high accuracies of 98.6% and 97.7% respectively.