Invited Lecture: Alina Donea, Monash University, Australia

Solar physics offers a uniquely rich and multi-scale environment for data science, combining high-resolution imaging, time-series analysis, and complex physical modelling. Modern missions, such as Solar Dynamics Observatory, continuously produce terabytes of data per day, including multi-wavelength images, Doppler velocity fields, and magnetic maps of the Sun. From a data science perspective, solar datasets are inherently high-dimensional, spatio-temporal, and physics-constrained, making them ideal for machine learning applications such as convolutional neural networks for image recognition, recurrent or transformer models for time-series forecasting, and unsupervised methods for pattern discovery. At the same time, they challenge purely data-driven approaches, as physical interpretability remains essential—for example, linking learned features to magnetic structures, energy transport, or wave dynamics.