Dr. Markus Stoye

Dr. Markus Stoye

I joined Reexen, where I lead a team to customize computer vision and audio task for a neural network accelerator developed by the company. The interesting challenges are the resource constraints and the errors induced bt the analog elements of the accelerator. Given company policies, no new results of my work is made public starting in April 2019. Thus, my newest research is not listed here. I am still affiliated with the Imperial College London, where I was working as Senior Scientist in the high energy physics group of Imperial College London and I am an Academic Fellow of the Data Science Institute. My research focus was big data science at the LHC collider at CERN, which produces petabytes of heterogeneous data every day. Together with my team, I have designed custom deep learning architectures, tools for mitigation of differences between real data and simulation, and interpretability. 

I co-organized the Machine Learning (IML) working group of the LHC experiments, which combined have about 10000 collaborators.

Recent presentations and seminars


  • “Likelihood-free inference with an improved cross-entropy estimator”, Poster at NeurIPS MLPS, Vancouver, December 20th

  • “AI in HEP, Colloqium, Max-Planck-Institute for Physics, Munich, Bavaria, Feb. 25th


  • Validation of DeepJetat CMS, ML fot jets, Fermi National Laboratory, Chicago, IL, USA, November 28th

  • Deep learning at the LHC, PPD seminar, Rutherford Appleton Lab., Oxfordshire, UK, July 25th

  • Deep learning at the LHC, HEP seminar, Imperial College London UK, July 24th

  • “Deep learning at CMS and Alice,” ICHEP, Seoul, Korea, July 5th

  • “Uncertainty mitigation in b-tagging,” IML (Inter-experiment LHC Machine Learning) workshop, CERN, Switzerland, 12th April

  • “DeepJet: jet tagging at the CMS experiment,” Big data science in astroparticle physics workshop, Aachen, Germany, Feb. 17th

  • “Machine Learning at the LHC,” ML for dark matter searches workshop, Leiden, Netherlands, Jan 17th


  • “Machine learning for jets at CMS,” ML in jets physics workshop, Berkley, USA, 11th December

  • “DeepJet: Generic physics object based jet multiclass classification for LHC experiments,” Deep learning for physics workshop at NIPS, LA, USA, 8th December

  • “DeepJet: performance and uncertainty mitigation”, IPPP seminar, Durham, UK, upcoming 16th November

  • “Machine learning applications in CMS,” CERN EP/IT data science seminar, CERN, 20th September

  • “Deep learning in jet reconstruction at CMS,” ACAT Seattle, USA, 21st August

  • “Convolutional neural networks for b-tagging,” DataScience@HEP, FNAL, 8th May

  • “DeepFlavour in CMS,” IML (Inter-experiment LHC Machine Learning) workshop, CERN, 21st March

  • LinkedIn Clean Grey