RESEARCH GROUPS

Area Architecture

Area Engineering

DEPARTMENTS

Feature-based machine learning and image-based deep learning for solar flare forecasting

keywords MACHINE LEARNING, DEEP LEARNING, OPTIMIZATION,

Reference persons EMMA PERRACCHIONE

Thesis type TESI SPERIMENTALE

Description Solar flares are the most explosive phenomena in the heliosphere; they originate from magnetically active regions (ARs) on the Sun but not all the active regions give rise to solar flares. The application of artificial intelligence techniques for their prediction is currently a hot topic in space weather. In the last decade machine and deep learning approaches have been obtaining an increasing interest in flare forecasting, thanks to flexible algorithms that may take as input physical features extracted from magnetograms of ARs, time series of features, or directly images of ARs, and videos of magnetograms of ARs. We investigate which are the advantages and limitations of machine learning techniques from physical features or deep learning techniques from images and we investigate hybrid approaches which can combine such techniques in order to obtain better predictions.

Deadline 27/01/2024      PROPONI LA TUA CANDIDATURA