Application of Equivariant Neural Networks to Boosted Top Tagging in CMS Open Data

Boosted top tagging has become an important classification task for measuring properties of the top quark in Large Hadron Collider (LHC) experiments. The landscape of this task, involving the classification of signal vs background in a large, complex dataset, is well suited for modern deep learning techniques which are capable of high performance on low-level data. We explore the application of a subset of these techniques, known as equivariant neural networks, on boosted top tagging using the ATLAS Top Tagging Open Data Set.