On the Viability of Optimized K-Means Clustering and Autoencoders for Detecting Network Anomalies

Document Type

Conference Proceeding

Source of Publication

2026 17th Student Research Conference on Applied Computing Src 2026

Publication Date

4-22-2026

Abstract

The technological evolution of the Internet of Things (IoT) has expanded the cyberattack landscape, resulting in new attack vectors. To keep up with these threats, we need to utilize technologies such as Artificial Intelligence (AI), particularly Machine Learning (ML) and Deep Learning (DL), to detect anomalous attacks. This study introduces PCA-IGKM, a novel ML algorithm that detects network anomalies in IoT networks. The study uses the IoT-TON-19 benchmark dataset and compares it with autoencoders, a DL algorithm. The findings demonstrate comparable results for PCA-IGKM to the AE-based model, with accuracy rates of 79% and 77%, respectively. Furthermore, this study marks the first implementation of autoencoders in the benchmark dataset, providing a baseline for future research. Although the results are promising, we still need to compare the novel model with other DL algorithms and test it across different datasets.

ISBN

[9798319510167]

Publisher

IEEE

Disciplines

Computer Sciences

Keywords

Anomaly Detection, Au-Toencoders, Deep Learning, Internet of Things, Intrusion Detection, K-means Clustering, Machine Learning

Scopus ID

105043176457

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS