Deep multi-task learning for stem water potential prediction : Actionable guidelines

Boje Deforce, Bart Baesens, Jan Diels, Pieter Janssens, Luis Bonet Pérez de León, Estefanía Serral Asensio

Abstract: Highlights•We propose actionable guidelines for predicting stem water potential effectively.•We leverage deep multi-task learning to forecast stem water potential in crops.•We achieve accurate results on two real-world datasets despite limited target data.•We analyze performance under various experimental setups in-depth.