Zoom:
https://cern.zoom.us/j/98857838063?pwd=U3hVUURjZ1J5SW50cnAwa0MxQ1pCQT09
In this talk I will introduce a recent ML approach based on point clouds named ABCNet. I'll show how this approach can be used for light quark and gluon separation, while also showing the performance for different flavours (uds) separately.
I will give a quick overview of jet flavor tagging for (SM+BSM) Higgs(-like) searches. I will cover standard "narrow-radius" jet tagging (inc. ML), boosted, "large-radius" jet tagging (inc. ML), the use of jet tagging in current and future searches, and common experimental considerations (detector design, simulation discrepancies, background modeling, systematics) for future studies.
We point out that the stringent lower bounds on the masses of new Higgs bosons crucially depend on the flavor structure of their Yukawa interactions. We show that these bounds can easily be evaded by the introduction of flavor-changing neutral currents in the Higgs sector, that are in agreement with low energy flavor constraints. As an illustration, we discuss the LHC phenomenology of a two Higgs doublet model with a Yukawa texture singling out the third family of quarks and leptons.