Anomaly Detection with Artificial Intelligence Methods – An Overview

Sebestyén, Pál György and Hangan, Lia-Anca and Czakó, Zoltán (2023) Anomaly Detection with Artificial Intelligence Methods – An Overview. MŰSZAKI TUDOMÁNYOS KÖZLEMÉNYEK (EN), 18. pp. 63-69. ISSN 2601-5773

Preview

Text MTK18-EN-12-Sebestyen.pdf - Published Version Download (465kB) | Preview

Abstract

Nowadays, more and more human activities depend on computer-based automated systems. Fully automat ed (robotized) production lines, energy distribution infrastructures and other urban services or environmen tal surveillance systems are just some examples of cyber-physical systems that depend entirely on automated control systems. In these cases a significant challenge is to identify abnormal behaviors of the supervised or controlled systems, in order to avoid malfunction or sometimes catastrophic events. Our main research goal was to evaluate the potential of adapting and using AI techniques in the field of anomaly detection. We also developed a platform, called AutomaticAI, which can help specialists in different domains to identify the best approaches to solve a given anomaly detection problem. The platform can select the best AI algorithm and parameter configuration for a given set of data containing normal and abnormal data. The tool was used successfully in a variety of domains, from cyber-physical systems to the medical domain.

Actions (login required)

Edit Item