Our ambition is to propose a general strategy for bringing physics information into the machine learning algorithms. The CORNERSTONE project is an interdisciplinary research project whose overall objectives are to build a data-driven computational prototype for the prediction of solar explosive events, to validate it through specific ‘science cases’. This ‘stand-alone’ application prototype will be based on machine/deep learning approaches trained with data contained in space and ground-based instrument archives in order to predict explosive events and identify their precursors. With reference to more specific objectives, CORNERSTONE will focus on the timely prediction of solar flares, Solar Energetic Particles (SEPs) and Coronal Mass Ejections (CMEs); identification of the data properties most involved in the prediction process; and automatic classification of active regions according to their probability of generating flares, SEPs or CMEs.
