LHC Physics Center

What I will be working on:

Among the many neural network architectures, graph-based models are particularly adapted to the particle collision data. Machine learning based event reconstruction and simulation algorithms have the potential of solving the increasing computation challenge we are facing in particle physics at the LHC. In particular generative models could offer orders of magnitude speed-up in simulation, and are getting sufficiently mature. Graph generative models are applicable at various levels of data representation ; at the particle flow candidate level for example. Event reconstruction with graph-based models can also be done at various levels of data processing, and in particular charged particle track reconstruction is amenable to such models.
As senior LPC AI fellow, I will pursue work on graph neural network models for event reconstruction and simulation, in particular towards integration to CMS production workflows. This work will find synergies with other machine learning activities at the LPC, Fermilab, CERN and world-wide.

My role in CMS past and present:

• 2020-current : CMS ML Group co-convener
• 2019-20: CMS ML Forum co-coordinator
• 2017-19: O&C DMWM co-convener
• 2015-17: O&C workflow management co-convener
• 2011-13: PdmV co-convener
• 2009-11: Reconstruction co-convener
• 2009: Tracking software co-coordinator
• 2008-09: SUSY software tools co-coordinator
• 2007-09: Muon HLT co-convener