Past observations of the Higgs boson produced with large mass quantum fluctuations had been severely limited in their sensitivity. In order to improve the sensitivity, a new paper by the ATLAS collaboration with participation from UMass Profs Rafael Coelho Lopes de Sa and Verena Martinez introduced a completely new way of interpreting their data. The method pioneered in a new result uses a neural network based technique called neural simulation-based inference (NSBI) to combat these issues. NSBI uses these neural networks to analyse each particle collision event individually, preserving more information and improving accuracy.
They were able to provide evidence for off-shell Higgs boson production with a significance of 2.5𝜎 (corresponding to a 99.38% likelihood), using events with four electrons or muons, compared to a significance of 0.8𝜎 using traditional methods in the same channel. The results mark an important step forward in understanding the Higgs boson as well as other high-energy particle physics phenomena. Two papers, one describing the NSBI method and another describing the new result, were published in the prestigious Report on Progress in Physics and highlighted in recent articles of Physics World.
https://physicsworld.com/a/probing-the-fundamental-nature-of-the-higgs-boson
https://physicsworld.com/a/neural-simulation-based-inference-techniques-at-the-lhc/