26TH INTERNATIONAL CONFERENCE ON COMPUTING IN HIGH ENERGY & NUCLEAR PHYSICS (CHEP2023)

High level reconstruction for e+e- Higgs factories with Machine Learning

Not scheduled

1h

Hampton Roads Ballroom and Foyer Area (Norfolk Waterside Marriott)

Poster Poster Poster Session

Speaker

Suehara, Taikan (Kyushu University)

Description

There exist emerging interests for e+e- Higgs factories in the context of FCCee feasibility study in Europe, ILC in Japan, and new proposals (C3, HELEN) at US. As an original developer of a widely-used software package of jet analysis including flavor tagging for e+e- linear collider studies, LCFIPlus (published at NIMA and arXiv:1506.08371), we are now developing DNN-based flavor tagging following original strategy of LCFIPlus. Based on the published results (arXiv:2101.11906, just accepted at NIMA), we present a trial to perform vertex reconstruction with RNN-based approach as well as more recent development for flavor tagging using GNN-based algorithm, accompanied with performance comparison with LCFIPlus. We are also working on DNN implementation of particle flow algorithm based on GravNet-based algorithm developed for CMS HGCal reconstruction, which will also be presented.

Presentation materials

Poster