Some magnetic features in active regions, related to strong solar flares, are considered as “anomaly” features in a machine learning algorithm. An unsupervised auto-encoder network has been trained to identify such anomalies and is used to predict occurrence of strong flares.
A sample of 32 flare events are analyzed to evaluate how these events agree with a flare-triggering model, which examines shear angles of large-scale magnetic field and small-scale dipole field during the flares’ precursor brightening.
A set of parameters that characterize the complexity and energy potential of solar active-regions is fed through several Machine Learning and conventional statistics algorithms to forecast solar flares.
A deep-learning method, Convolutional Neural Network, is developed to use the HMI’s line-of-sight magnetic field to forecast solar flares.