Professor Luís Miguel Pires publishes a study in MDPI Electronics on intelligent anomaly detection in IoT systems, reinforcing IPLUSO’s research in artificial intelligence and data analytics
Professor Luís Miguel Pires recently published the scientific article "An Interpretable Agent-Assisted Pipeline for Statistical Anomaly Detection in IoT Temperature Time Series" in the journal MDPI Electronics, volume 15, issue 9, article 1840.
The study proposes an innovative approach to anomaly detection in IoT sensor time series, combining interpretable statistical methods with intelligent agent-assisted mechanisms, with the aim of enhancing robustness, adaptability, and trust in IoT systems applied to environmental and industrial monitoring.
The research explores statistical filters such as Hampel, IQR, and Z-score, integrated into an intelligent pipeline capable of supporting the adaptive selection of detection strategies in dynamic and heterogeneous scenarios. The main contributions include:
- Development of an interpretable pipeline for anomaly detection in IoT data;
- Integration of agent-based assistance for dynamic adaptation of detection methods;
- Experimental validation across multiple synthetic scenarios and real sensor datasets;
- Contribution to more reliable, explainable, and energy-efficient IoT systems.
This work positions itself at the intersection of Internet of Things, explainable machine learning, and adaptive autonomous systems, addressing emerging challenges related to edge intelligence and the reliability of smart sensors.
According to the author, the study opens future perspectives for integration with Reinforcement Learning, multi-agent systems, and distributed architectures for autonomous monitoring in critical applications. This publication strengthens IPLUSO’s scientific activity in the fields of intelligent IoT, data analytics, and applied artificial intelligence, contributing to the institution’s international research visibility.
Read the Article Here
