Self-attention for The Image Completion Task on the Calorimeter Data in High Energy Physics

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21340%2F23%3A00369788" target="_blank" >RIV/68407700:21340/23:00369788 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Self-attention for The Image Completion Task on the Calorimeter Data in High Energy Physics

Popis výsledku v původním jazyce

Detector simulations are an important component of the research in high energy physics. However, the traditional Mote Carlo-based simulation tools are computationally intensive and finding a faster alternative is necessary. Development of the deep learning models for detector simulations has been of interest to the HEP Community with main focus on the MLPs and CNNs. However, these models are usually limited to one very specific detector and particle setup. In the aim of having a more general model, we explore the self-attention mechanism for an image completion task on the calorimeter data. Inspired by the vision transformer (ViT), we treat the calorimeter showers as images and use a transformer for masked sequence modelling. We also leverage the graph convolutional layer for embedding of the showers.

Název v anglickém jazyce

Self-attention for The Image Completion Task on the Calorimeter Data in High Energy Physics

Popis výsledku anglicky

Detector simulations are an important component of the research in high energy physics. However, the traditional Mote Carlo-based simulation tools are computationally intensive and finding a faster alternative is necessary. Development of the deep learning models for detector simulations has been of interest to the HEP Community with main focus on the MLPs and CNNs. However, these models are usually limited to one very specific detector and particle setup. In the aim of having a more general model, we explore the self-attention mechanism for an image completion task on the calorimeter data. Inspired by the vision transformer (ViT), we treat the calorimeter showers as images and use a transformer for masked sequence modelling. We also leverage the graph convolutional layer for embedding of the showers.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10303 - Particles and field physics

Projekt

<a href="/cs/project/LM2023061" target="_blank" >LM2023061: Výzkumná infrastruktura pro experimenty ve Fermilab</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

SPMS 2022/23 Stochastic and Physical Monitoring Systems, Proceedings of the international conferences

ISBN

978-80-01-07250-9

ISSN

—

e-ISSN

—

Počet stran výsledku

8

Strana od-do

45-52

Název nakladatele

České vysoké učení technické v Praze

Místo vydání

Praha

Místo konání akce

Sloup v Čechách

Datum konání akce

26. 6. 2023

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—