Kick-off: Workshop on task “Machine Learning” held in Dresden

On 27 and 28 February, a workshop to kick-off Task 5.4 of the THRILL project brought together laser operators and experts from ELI beamlines, GSI, HIJ, HZDR and LULI to share their approaches to laser-operations data acquisition and laser control. The workshop’s participants also took the opportunity to discuss where Machine Learning (ML) could best be applied to help solve major open issues in laser beam quality and stability.

On both days, keynotes by Vincent Bagnoud (GSI), Ulrich Schramm (HZDR) and Thomas Cowan (HZDR/HIBEF) provided overviews of the science done at their facilities and gave structure to the many aspects of beam stability and impact.

Technical presentations about the laser control systems’ capabilities for data collection and digital system control met talks about data exchange formats and data management, leading to discussions on the opportunities and challenges for data-hungry ML methods. This enabled us to plan the next steps to tackle concrete problems now, and to work on the data pipelines to approach more problems later. Talks on laser beam stabilization techniques, some involving ML methods for optimization and time-series prediction, currently employed or envisioned at the various facilities sparked lively discussions, exchange of experiences, and identified avenues for future collaboration between groups.

The workshop was capped-off by tours of the high-power lasers DRACO and PENELOPE operated at the HZDR campus in Dresden Rossendorf, which are contributing to the work done within the THRILL consortium.

Text: Jeffrey Kelling

DRACO control room

Subscribe to our newsletter

For regular updates on THRILL, sign up for our newsletter.