by Andrey Ustyuzhanin

Founded in 2007 by Russia’s leading web search provider and one of its largest IT companies, the Yandex School of Data Analysis (YSDA) has been offering the knowledge, experience, skills and insights of its teachers, students and graduates to the global high-energy physics community since 2011.

With its free Master’s-level program in computer science and data analysis, the Yandex School of Data Analysis brings together accomplished scientists, researchers, and engineers, who share their knowledge and experience with graduates in software engineering, mathematics, information retrieval and related fields.

Designed by Yandex to create and nurture a pool of talent primarily for the company’s own needs, the School now welcomes students and interns from all over the world who take this opportunity to hone and develop their skills, pursue their interests and apply their ideas in the real-life environment provided by a major technology company.

unnamedLocated at Yandex’s headquarters in the heart of Moscow, in addition to offering hands-on experience in developing breakthrough technologies and end-user products, the School gives its students and interns an opportunity to join lectures and seminars by legendary scientists (Albert Shiryaev reads courses on stochastics and probability models), as well as new-generation contributors to data science (Konstantin Vorontsov with his methods in machine learning and Victor Lempitsky sharing all he knows about deep learning).

The School’s joint programs with Russia’s leading universities, including the Moscow Institute of Physics and Technology, the Lomonosov Moscow State University, and the Higher School of Economics, create a unique multidisciplinary platform for sharing knowledge and generating ideas.

The School’s faculty members, active researchers and experienced engineers also contribute to a number of specialist contests and challenges, including the ‘Flavours of Physics’ contest co-organized with LHCb, University of Zurich and University of Warwick in 2015.

The YSDA is also one of the organisers of LHCb’s annual Summer School on Machine Learning in High Energy Physics, which took place last year in Saint Petersburg and will be hosted by Lund University in Sweden this year.

Following a long-term collaboration of its founding company, Yandex, with LHCb through CERN’s openlab partnership program, the YSDA joined LHCb as an associate member in 2014, and became its full member in 2015. The School’s specialists provide continuous support to LHCb’s EventIndex (which will soon be joined by the forthcoming EventFilter) and develop new tools and services specifically for the needs of CERN researchers.

The YSDA’s collaboration with LHCb has already resulted in an optimization of the speed of its topological trigger by using Yandex’s machine learning algorithm, MatrixNet, which improved the signal efficiency in Run-II vs Run-I by up to 60%. Optimization of the experiment’s grid data storage by the YSDA specialists resulted in saving up to 40% of disk space.

The School’s collaboration with LHCb also produced the Reproducible Experiment Platform(REP), a software infrastructure that allows research teams to run computational experiments on shared big datasets, including those in high-energy physics.

Another product of the YSDA-LHCb collaboration is a service for managing Jupyter-based research environments, everware, which, among other things, allows performing studies, such as gravitational wave analysis, or CP-violation, or a J/psi mass analysis.

The range of products developed by the Yandex School of Data Analysis for LHCb includes tools and services for monitoring data quality and predicting anomalies in the enormous amounts of data generated in the experiment. The YSDA continues seeking new innovative ways of applying data analysis and machine learning in high-energy physics, and welcomes research visitors to work on its projects in Moscow.