Domain Adaptation For H and Z Jet Distributions.

Jet identification (or "tagging") is one of the most important task in collider physics. However, the performance of the jet tagging algorithms depend on many parameters. For instance, an algorithm trained jets using jets from one source, can be shifted in a way that the same trained model does not perform well on another source. In this report, unsupervised adversarial domain adaptation techniques were implemented to the ParticleNet model using jet data sets from Higgs or Z bosons decays with a qualitative, and quantitative illustration of the shift.