From Jet tagging to event classification using Transformer

You-Ying Li^1*, Zheng-Gang Chen¹, Yi-An Chen¹, Kai-Feng Chen¹

¹Department of Physics, National Taiwan University, Taipei, Taiwan

* Presenter:You-Ying Li, email:yyaustinli@ntu.edu.tw

Deep learning architectures that use particle-level features have significantly advanced jet tagging applications, playing a critical role in many important HEP analyses, such as b-tagging techniques in Higgs-to-bottom-quark searches. In particular, Transformer architectures, which capture full correlations between particles within jets, have demonstrated extraordinary performance as one of the next-generation machine learning candidates. This could excite curiosity: can such Transformers, using particle-level features, also enhance event classification tasks beyond traditional high-level jet features? In this talk, we will present an experiment focused on event classification for Higgs production, specifically distinguishing between Vector Boson Fusion (VBF) and gluon fusion processes as an illustrative example.

Keywords: Deep learning, Transformer, Event classification, Jet Tagging