WG1 · Machine Learning Applications for Collider Physics

From fast inference to anomaly detection, WG1 turns advanced ML research into tangible collider capabilities.

Working Group 1

WG1 · Machine Learning Applications for Collider Physics

WG1 transforms cutting-edge machine learning research into deployable models entering into the experimental workflow for triggering, reconstruction and analysis while delivering robust and improved performance.

Co-Leaders: Dr Valentina Vecchio, Prof Pietro Vischia

  • Advanced ML architectures for detector reconstruction and precision triggering.
  • Unsupervised and weakly supervised anomaly detection to uncover unexpected signals.
  • Physics-informed networks with quantified uncertainties for high-stakes analyses.
Dr Valentina Vecchio

WG1 Co-Leader

Dr Valentina Vecchio

valentina.vecchio@cern.ch

University of Manchester
Oxford Road
Manchester M13 9PL, UK

0000-0002-1351-6757

INSPIRE profile 1505521

Prof Pietro Vischia

WG1 Co-Leader

Prof Pietro Vischia

pietro.vischia@gmail.com

Universidad de Oviedo
Leopoldo Calvo Sotelo 18
33007 Oviedo, Spain

0000-0002-7088-8557

INSPIRE profile 1054943

Current priorities

WG1 works closely with experts from the experimental collaborations to validate the ML models, ensuring transparent performance, reproducibility, and alignment with MC-approved milestones.

Key outputs
  • Release open benchmarks tailored to collider ML challenges.
  • Prototype low-latency inference chains ready for online selection.
  • Publish guidelines for uncertainty-aware ML workflows within the Action.
Collaboration touchpoints
  • Coordinate with WG3 to align ML pipelines with theory-driven use cases.
  • Work with WG4 on scalable deployment across shared infrastructure.
  • Support WG5 in packaging ML knowledge into training assets.