About
I am a postdoctoral researcher at the Center for Plasma Astrophysics (CmPA) at KU Leuven, where I develop a space-weather forecasting framework using machine learning (ML) and deep learning methods. I build data-processing pipelines that combine diverse analysis techniques with ML models—such as β-VAE, k-NN, and LSTMs—for time series, spectra, and images.
My research centers on multi-wavelength, information-rich analyses of solar active regions. I am particularly interested in large, magnetically complex regions, their dynamical evolution, and their flaring potential and eruption precursors. The distinction between eruptive and confined flares remains an open question and an active area of research. To address it, we leverage both data-driven (machine-learning-based features) and physics-based parameterisation of the available observations. Success on this front could enable the integration of these insights into on-board diagnostics and analysis for the next generation of Solar System explorers.
Contact
ekaterina.dineva@kuleuven.be · GitHub · LinkedIn · Bluesky 🦋