Future of hadronically decaying tau identification and reconstruction

Tau leptons play a crucial role in many HEP analyses, thereby signifying the importance of an accurate 𝜏h identification and reconstruction. Following the direction of the advances made in ML applications for jet tagging, we study the usability of transformer based models for the three main tasks in 𝜏h reconstruction – tagging, kinematic reconstruction and decay mode classification. For evaluating said models, we created a new publicly available full-sim dataset, Fu𝜏ure, that also aims to provide a solid grounds for the development of new 𝜏h reconstruction algorithms. With an outlook for the future, we further explore the possibility of using foundation models for the same task.

Speaker: dr. Laurits Tani

When: 25^th November at 3.30 pm
Where: Aula Multimediale – Dipartimento Interateneo di Fisica – Bari