How deep learning is complementing deep thinking in ATLAS

Abstract

AbstractATLAS collaboration uses machine learning (ML) algorithms in many different ways in its physics programme, starting from object reconstruction, simulation of calorimeter showers, signal to background discrimination in searches and measurements, tagging jets based on their origin and so on. Anomaly detection (AD) techniques are also gaining popularity where they are used to find hidden patterns in the data, with lesser dependence on simulated samples as in the case of supervised learning-based methods. ML methods used in detector simulation and in jet tagging in ATLAS will be discussed, along with four searches using ML/AD techniques.