Search for H→aa→bbbb boosted events in 2018 CMS data using neural networks

This talk will present the results and methods of a search for pseudoscalar a, predicted under two-higgs doublet extensions to the standard model, at a 20 GeV theoretical mass point. The decay path H->aa->bbbb in highly boosted jets was specifically targeted, and a binary classification neural network was developed to perform the search using data collected by the CMS experiment in 2018. Upper limits on the existence of a will be presented, and much of the talk will be dedicated to covering neural network techniques (including the use of focal and distance-correlation loss) and lessons in network development learned along the way.