Multiscale Dynamics Seminar Ajay Tiwari

Title: Machine Learning in Space Weather

Abstract: The ever increasing popularity and recent advancements in machine learning (ML) techniques have led to a variety of new tools for resolving traditional and emerging challenging problems more powerfully from data-driven perspectives. There have been various attempts to catalogue Coronal Mass Ejections (CME) observations, in this work we benchmark four of the more popular CME databases for CME time-of-arrival prediction using machine learning methods. We identify the important features in the various databases and quantify the performance of machine learning methods. To this end we have also created an interactive CME ML playground where the databases can be accessed and trained on the cloud.