The MLISE conference provides an international forum for researchers and practitioners to exchange ideas, present scientific advances and discuss technological applications in the fields of machine learning and intelligent systems engineering. Key topics include deep learning, data mining, computer vision and natural language processing.
At the 2026 edition, held in Naples from 28 to 31 May, Professor Francesco Flammini delivered a plenary keynote entitled Safe Perception for Trustworthy Autonomy: From Robust Sensor Fusion to Run-Time Adaptation. He also received the Best Paper Award for the paper Towards Energy-Efficient Federated Anomaly Detection in Dynamic Environments, developed in collaboration with researchers from the University of Naples Federico II and the IMT School for Advanced Studies Lucca. The paper was selected as the best among more than 100 submissions.
The research addresses a critical challenge for distributed Internet of Things (IoT) systems: reliably detecting anomalies and faults even as operating conditions evolve over time. This issue, known as concept drift, occurs when patterns that are considered normal gradually change. For example, a sensor installed in a manufacturing plant may produce different readings as machinery ages or production processes are modified. As a result, data that appears normal today may become anomalous in the future.
To address this challenge, the researchers developed an approach that enables devices to learn collaboratively without directly sharing the data they collect or continuously transmitting large volumes of information to a central server.
“When operating conditions change, monitoring systems may generate false alarms or fail to detect actual issues,” explains Flammini. “Our solution continuously tracks how conditions evolve and automatically updates its analytical models whenever significant changes in normal behaviour are detected. Each device processes data locally while contributing to a shared model, reducing both data traffic and computational requirements.”
In the coming years, technologies of this kind could help improve the safety, reliability and efficiency of monitoring systems in strategic sectors such as advanced manufacturing, critical infrastructures, smart mobility and energy networks.