BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:GUNTAM: A Transformer for track seed reconstruction
DTSTART:20261023T073000Z
DTEND:20261023T083000Z
DTSTAMP:20261007T090100Z
UID:indico-event-40615@indico.in2p3.fr
DESCRIPTION:Speakers: Corentin Allaire (IJCLab\, Université Paris-Saclay\
 , CNRS/IN2P3)\n\nReconstructing particle trajectories is a significant cha
 llenge in most particle physics experiments and a major consumer of CPU re
 sources. It can typically be divided into three steps: seeding\, track fin
 ding\, and track fitting. Seeding involves identifying potential trajector
 y candidates\, while track finding entails associating detected hits with 
 the corresponding particle. Finally\, track fitting focuses on reconstruct
 ing the parameters of the trajectory. In classical tracking\, most of the 
 intensive computational workload stems from the seed identification proces
 s\, while track finding is well understood. To address the complexity of s
 eed reconstruction\, we introduce GUNTAM\, a transformer-based algorithm f
 or seed reconstruction. This algorithm takes as input all the 3D space poi
 nts in different phi bins through the detector and returns\, for each\, a 
 list of possible seeds. This network exploits the Transformers’ attentio
 n matrix's natural ability to predict correlations between objects. With p
 roper training\, it is possible to turn from such a matrix into a graph ad
 jacency matrix connecting together hits from the same particle and thus re
 constructing with extremly high performance the seed corresponding to the 
 original particles.\n\nhttps://indico.in2p3.fr/event/40615/
LOCATION:Amphi G.Besse (Subatech IMT Atlantique)
URL:https://indico.in2p3.fr/event/40615/
END:VEVENT
END:VCALENDAR