Spåtind 2023 - Nordic Conference on Particle Physics

Using computer-vision inspired techniques for end-to-end event classification on ATLAS data

5 Jan 2023, 20:50

15m

Fefor Høifjellshotell - Norge

Contributed Talk ML and AI for physics Contributed Talks IV

Speaker

Aurora Singstad Grefsrud (University of Bergen (NO))

Description

The ATLAS detector is, in its most abstract form, a cylindrical camera. It captures the energy deposits left in the detector material from particles produced in high-energy particle collision events. Traditional methods of analysing this data rely on the reconstruction of the particle collision. This process transforms the high-dimensional, low-level data from the detector into lower-dimensional, high-level data. By treating the low-level energy deposit data as part of a panoramic (360 degrees) image, we aim to skip the reconstruction event and classify the event based on this “energy fingerprint” using machine learning algorithms specifically designed for image recognition.

Presentation materials

Spaatind presentation (1).pdf

Spaatind presentation.pdf