Using Advanced Machine Learning Techniques to Study Poorly Modeled Processes in <em>pp</em> Collisions with the ATLAS Detector

Abstract

In high energy particle physics, measurements are made to improve and test our models. Some processes are difficult or impossible to model with current capabilities. For some, this means that one would estimate such processes via data-driven techniques or use less-than-ideal modeling during the measurement. The aim of the research presented in this paper is to explore advanced machine learning techniques to deal with hard-to-model or unmodeled processes in analyses using the ATLAS detector. The two tackled problems addressed in this document are hadronically decaying tau-lepton identification and generating a sensitive variable for a WWbb measurement that enhances tW and ttbar interference.