Automatic Detection of Occulted Hard X-Ray Flares Using Deep-Learning Methods

We present a concept for machine-learning classification of hard X-ray (HXR) emissions from solar flares observed by the Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI), identifying that are either occulted limb or located on disk. Although HXR observations important particle-acceleration studies, data analyses past were time consuming and required specialized expertise. Machine-learning techniques promising this situation, we constructed sample model to demonstrate using deep-learning technique. Input spectrograms easily produced RHESSI data. The can detect without need image reconstruction nor visual inspection experts. A technique convolutional neural networks was used in regarding input as images. Our achieved accuracy better than 90 %, ability application method event screening an alert successfully demonstrated.