Identification of Electron Diffusion Regions with a Machine Learning Approach on MMS Data at the Earth's Magnetopause

This article presents 18 magnetic reconnection electron diffusion region (EDR) candidates found using a neural network algorithm with the Magnetospheric Multiscale Mission phase 1a data at Earth's dayside magnetopause. These new are compared to 32 previously reported EDRs listed in Webster et al. (2018), https://doi.org/10.1029/2018ja025245, which constitute training database of our algorithm. One main parameters used is scalar quantity called “MeanRL” based on asymmetry velocity distribution function and better identifies agyrotropy plane perpendicular field. In light EDR found, we discuss analyze sign energy dissipation during process distinction between inner outer EDRs, 40% showing negative or oscillating dissipation. We also present details one identified candidates.