Development of novel experimental techniques to improve the understanding of the Higgs sector by the ATLAS experiment

With the full Run-2 (2015-2018) proton-proton collision data collected by the ATLAS detector at the Large Hadron Collider, precise measurements of Higgs boson properties in an array of production and decay modes are now possible. To maximise the scientific value of the recorded data, novel experimental techniques were developed. The following article reviews a represen- tative selection of such techniques, which includes: multi-class machine learning classification optimisation algorithms, experimental uncertainty regression, input variable invariant adversarial neural networks, object embedding, and multi-dimensional likelihood re-weighting techniques designed to maximise the statistical precision of Monte Carlo predictions.