School speakers

Victor Kitov (MSU, Russia)

PhD from Moscow State University (2004-2008), Computer Science
MSc in economics from New economic school, Moscow
Scientific interest in to statistics, time series analysis and machine learning.
Has prepared and given courses on Machine Learning at

Skolkovo institute of science and technology (Moscow) in 2014
London Imperial College (London) in 2015.

Experience:

Development of a system for automatic extraction of topics from text collections, Russian academy of sciences.

Development of traffic prediction algorithms, Huawei technologies research center (owns 1 patent).
Development of an algorithm for optimal distribution of goods for warehouses, “Svaznoy” retail network,
Development of demand prediction algorithms for retail networks, “Forecsys” company.

Alexey Rogozhnikov (Yandex School of Data Analysis, Yandex Data Factory, Higher School of Economics, Russia)

PhD from Moscow State University (2012-2015), Computer Science
Graduated from Yandex School of Data Analysis (2014)
Prepared and given Machine Learning course in Imperial College, London, 2015
Currently works on YSDA-CERN projects as data scientist at Yandex (LHCb, SHiP)

Scientific interest: improvements of machine learning algorithms for solving field-specific problems (HEP, information retrieval)

Mika Anton Vesterinen (Heidelberg University, Germany)

PhD from the University of Manchester (2007-2011) on electroweak physics at D0, Fermilab

Rutherglen thesis prize 2011
Springer thesis prize 2012
University of Manchester PG student of the year in science and engineering 2012

CERN fellow (2011-2013) on LHCb
Postdoc with Heidelberg university (2013–), LHCb
Alexander von Humboldt fellow (2015–), with Heidelberg, LHCb
LHCb Semileptonic WG convenor from 01/2014–04/2016
LHCb Trigger deputy project leader from 10/2014–

Mika’s talk will cover the challenge of triggering on complex signatures at hadron collider experiments. What are the typical signals and sources of background? Figures of merit in the trigger performance beyond raw efficiency (biases causing systematic uncertainties etc).

Patrick Owen (Imperial College London, UK)

PhD in particle physics from Imperial College London (2010-2014)

Lead author of 4 physics publications which used machine learning techniques.

Postdoc at Imperial College London (2014-) on the LHCb experiment
Physics interests:

Rare decays of B mesons.
Precision measurements of so-called ‘semi-leptonic’ decays.

Patrick’s talk will discuss the challenges of searching for physics signatures with a large amount of background, and how machine learning techniques are being used in the LHCb experiment to improve precision of physics measurements. The talk will focus on some of the specific issues encountered in high energy physics.

Lesya Shchutska (University of Florida, USA)

MSc from Moscow Institute of Physics and Technology (2008) on LHCb
PhD from Ecole Polytechnique Federale de Lausanne (2008-2012) on PEBS
CHIPP prize as the best PhD student in theoretical and experimental particle physics in Switzerland (2011)
Postdoc at University of Florida (2012-), CMS
carrying out and leading SUSY searches with multileptonic signatures
CMS Leptonic SUSY WG convener (2015)
tau->3mu feasibility studies and experiment design with SHiP facility (2014-)

The lecture will address the CMS experience with MVA vs cut-and-count techniquesin the SM model processes searches and measurement, and with MVA applicationin searches for specific SUSY scenarios, and also will include a discussion on the placeof advanced techniques in the general new physics searches.

Josh Bendavid (CERN fellow, Switzerland)

PhD from Massachusetts Institute of Technology (2007-2012) on CMS
CERN Fellow (2012-2015)
Close involvement in Higgs to di-photon search, observation, and measurements in CMS, and the B_s→ μ⁺μ^- search in CMS and combination with LHCb
CMS Monte Carlo Generator convener (2014-2015)

The talk will cover the role of multivariate techniques in the Higgs search, discovery and measurements, from the level of physics object reconstruction and identification, all the way through to the extraction of final results and the interplay with statistical methods and systematic uncertainties.