The day in conversation for 2026-10-31

discussion

Transcript

The discussion: Kai: The embedding strategies introduced in that HAXAD paper seem to shift the sensitivity curve significantly, especially when looking for subtle Higgs signatures that might have been missed by standard supervised learning approaches; I wonder if those new embeddings introduce a bias towards specific kinematic regions we haven't properly explored yet. Mira: That sensitivity gain is compelling, but I’m more interested in how those ML techniques handle the inherent backgrounds in BESIII data when we look at charmed baryon decays; specifically, does the deep learning approach you mentioned manage to robustly disentangle polarization effects from known decay modes without introducing spurious correlations? Lev: If the polarization measurements at BESIII are clean, that implies a level of control over hadronic structure that’s hard to achieve across all experiments; when we look at those new ML signatures, what is the quantifiable uncertainty budget on the signal versus the systematic uncertainty derived from model assumptions in those embedding frameworks?

More episodes

← Home