Dresden 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

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

AKBP: Arbeitskreis Beschleunigerphysik

AKBP 14: Focus session: Machine Learning

AKBP 14.3: Hauptvortrag

Donnerstag, 19. März 2020, 10:30–11:00, MOL 213

Towards Reinforcement Learning Based Optimization at the Light Source BESSY II — •Luis Vera Ramirez — luis.vera_ramirez@helmholtz-berlin.de

The incorporation of Machine Learning tools in order to improve the performance and the experimental setups at the large-scale user facility BESSY II is an important part of the next years' roadmap at the Helmholtz-Zentrum Berlin. In this talk, we will focus on several use cases based on Reinforcement Learning (RL) based on simulations as well as experiments with real machine data. These experiments are currently being developed by both the beamline and the accelerator groups and include, among others, the optimization of booster current and injection efficiency (already carried at out the machine) as well as first RL-approaches regarding automatic parameter tuning for beamline raytracing.