Abstract: We propose an anomaly detector-agnostic framework to exploit heterogeneous and multidimensional streams in industrial predictive maintenance, with the main objective of detecting early data ...