A hybrid supervised/unsupervised machine learning approach to solar flare prediction

This work addresses solar flare forecasting for space weather applications, but it appears incremental as it builds on existing methods without major breakthroughs.

The authors tackled solar flare prediction by combining supervised regularization for feature importance with unsupervised clustering for binary classification, validating their hybrid approach on NOAA SWPC data.