Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: Applications in Particle and Astroparticle Physics

AKPIK 2.4: Vortrag

Dienstag, 21. März 2023, 17:45–18:00, ZEU/0118

ProGamer: PROgressively Growing Adversarial Modified (transformer-)Encoder Refinement — •Benno Käch, Isabell Melzer-Pellmann, and Dirk Krücker — Deutsches Elektronen-Synchrotron DESY, Hamburg, Germany

Machine learning-based data generation has become a major research topic in particle physics due to the computational challenges posed by current Monte Carlo simulation approaches for future colliders, which will have significantly higher luminosity. The generation of collider data is similar to point cloud generation, but it is more difficult because of the complex correlations that need to be accurately modeled between the points. A refinement model consisting of normalising flows and transformer encoders is presented. The normalising flow is 3-dimensional, meaning that the generated particle cloud consists of independent and identically distributed objects. This output is then refined by a transformer encoder, which is adversarially trained against another transformer encoder discriminator/critic. As the model is able to produce an arbitrary number of particles, a progressively growing point cloud can be produced.