Click here to download the session cfp.

The massive digitalization of industrial assets carried out in the last decade, allows now the deployment of industrial cyber-physical systems and IIoT architectures including advanced analytics capabilities to improve decision-making processes. In this regard, machine learning (ML) is being included as part of industrial maintenance procedures to improve anomaly detection, fault diagnosis and remaining useful lifetime prediction over assets. However, two main challenges have to be faced for an effective deployment of such ML-based procedures: first, the scalability, that is, hyperparameters selection procedures and interpretable training methods avoiding overfitted algorithms, and, second the reliability, that is, continuous learning approaches avoiding deviations between the supervision model and the operational drifts during the assets useful life.