Report

PUBDB-2025-04910

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Tracking in Dense Environments with Transformers

ATLAS Collaboration (Collaboration author)DESY*

2025

This record in other databases:  

Please use a persistent id in citations: doi:10.3204/PUBDB-2025-04910

Report No.: ATL-PHYS-PUB-2025-045

Abstract: This work presents a novel application of machine learning to the pattern-recognition stage of charged-particle reconstruction, enabling learned hit-to-track association within dense environments, such as the cores of high-pTjets. Our Transformer-based architecture is based on a MaskFormer model that jointly optimises hit assignments and the estimation of the charged particles' properties. Trained and evaluated in dense environments the model delivers up to a 30% improvement in track-reconstruction efficiency over the standard ATLAS reconstruction when local particle density makes conventional reconstruction most challenging.Scientific contact person Juste Rozas, Aurelio, (aurelio.juste.rozas@cern.ch)

---

Contributing Institute(s):

1. LHC/ATLAS Experiment (ATLAS)

Research Program(s):

1. 611 - Fundamental Particles and Forces (POF4-611) (POF4-611)

Experiment(s):

1. LHC: ATLAS

Appears in the scientific report 2025

---

Database coverage:
;

---