Machine learning techniques in particle physics

The classification of charged particle jets according to their pattern they leave in the calorimetric cells of the ALICE (CERN) detector can be approached by specifically tailored convolutional neural network if one treats traces of the jets as images. One particularly important target of our jet ML models is the signal-to-background ratio which is learned using experimental data obtained from known jets embedded in background. Boosted Decision Trees is an appropriate shallow ML approach for particle identification and it is used in models often trained using realistic Monte Carlo simulations.

We disseminate our results and knowledge by giving research talks and seminar talks as well as by teaching machine learning university courses.

Machine learning in statistical and quantum physics

Our interests address learning algorithms which we use to train physically relevant models of quantum systems such as Boltzmann machines and yield new classes of meaningful solutions of ensembles of interacting particles. Quantum machine learning algorithms involving reinforcement learning techniques help us extract prominent and very perspective testbed for future applications of quantum computers in optimizing objective functions of many-body systems.

We teach university courses  in quantum computations.

Group members

Dr. Peter Hristov joined CERN in 2000 as a member of the ALICE collaboration.He worked on the software for offline data processing and as the ALICE programme librarian. He is currently leader of the Software Development section of the Experimental Physics department. Prior to joining CERN, he was a research physicist and participated successively in the EXCHARM experiment at the U-70 facility at Protvino, Russia, and in the NA48 experiment at the SPS facility at CERN. Dr. Hristov obtained a PhD in Particle Physics in 1999 from the Joint Institute for Nuclear Research, Dubna, Russia.

Dr. Vesselin G. Gueorguiev is an established researcher with more than 50 publications (with 753 citations), including 1 monograph. His research interests span from group theoretical methods in physics to General Relativity and Cosmology.

The Research Statement of Dr. Gueorguiev can be found here.

Dr. Stoyan Mishev is a researcher with more than 20 publications in journals with high impact factor in which he is a major author. His research interest are in the many-body quantum theory with applications to atomic nuclei, nuclear astrophysics as well as machine learning. Dr. Mishev defended his PhD thesis in JINR (Dubna, Russia) in 2011 where he worked until January 2017. His major contributions are in the particle correlations of many-nucleon systems. Currently he is a lecturer in physics and computer science at New Bulgarian University and a scientific associate at the Institute for Advanced Physical Studies.