Comparative Analysis of Machine Learning and Deep Learning Models for Attack Detection in IoT and Conventional Networks

The rapid dissemination of IoT devices has brought many benefits, but has also introduced significant cyber security challenges, especially due to their weaknesses for network infiltration. This paper presents a comparative analysis of machine learning and deep learning models to detect network infiltration using IoT-23 and NSL-KDD dataset. The purpose of the study is to assess the performance of the random forest (RF) and Graph Convolutional Network (GCN) models for malicious activity identification in the IoT and traditional network environment. In particular, random forest model showed excellent accuracy, scored 91% on IoT-23 dataset and 99.91% on NSL-KDD Dataset. A good performance was also given by the GCN model, which acquired an accuracy of 71.87% and 96.6% on the respective dataset using the ability to catch complex graph-composed relationships. The results obtained displays the effectiveness of the use of traditional as well as deep learning techniques to detect infiltration, which further provides insight into the IoT ecosystem and their viability to secure the traditional network. Results provide a foundation for the construction of strong, scalable infiltration detection systems and implications for improving cyber security in diverse environments.