Ilya Segalovich Scholarships Awarded on the Fifth Anniversary of HSE’s Faculty of Computer Science

As part of the HSE Faculty of Computer Science fifth anniversary celebration at Mercury Moscow City Tower, Ilya Segalovich Scholarships were awarded.

The first scholarships were awarded in 2015, on the first anniversary of the Faculty. Over the years, the Faculty has become one of the most recognized faculties not only at HSE, but among Russia’s leading IT faculties.

The HSE Faculty of Computer Science was created in collaboration with Yandex, based on Stanford University and EPFL’s example. Today, the Faculty has increased the number of its partner projects with Yandex, and it also cooperates with Samsung, SAS, CERN and Sberbank. Since the Faculty is popular among mathematic and programming test winners, the number and level of undergraduate and Master’s students is growing annually, and the level of education is growing accordingly.

The Faculty of Computer Science today

1721

students

3

undergraduate programmes (including an English-taught double degree programme with the University of London)

6

master’s programmes

9

research laboratories

9

Russian Science Foundation grants

Supporting young researchers is a one of the Faculty’s top priorities, and the Ilya Segalovich Scholarship is one of the ways to motivate the best students who carry out projects and research. The scholarship awards undergraduate, master’s and doctoral students a monthly stipend of 30,000, 35,000, and 40,000 rubles, respectively, for one year.

This year, 16 students have been awarded the scholarship: Nikita Veselko, Filipp Gribov, Nikita Kalinin, Daniil Nikolenko, Mikhail Nosovsky, Irina Ponamareve, Vladimir Smurygin, Polina Tarantsova, Daniil Tyapkin, Anastasia Chistopolskaya, Artem Gadetsky, Ekaterina Kozlova, Alexander Lazarenko, Vladislav Belavin, Olga Gerasimova, and Roman Nesterov.

Igor Agamirzian, Vice President of HSE, Maria Yudkevich, HSE Vice Rector, Sergey Roshchin, HSE Vice Rector, Arutyun Avetisyan, Director of the Ivannikov Institute for System Programming, Andrey Sobolevsky, Director of Kharkevich Institute for Information Transmission Problems, Mikhail Levin, Deputy Head at Yandex Banner System Service, as well as representatives of Sberbank, Samsung, 1C and other faculty partners came to congratulate the Faculty of Computer Science and award the winners.

This year, the first Ilya Segalovich Award for contribution to the development of computer science was delivered, which has also been founded by Yandex

2019 winners spoke with the HSE News Service about their projects and research interests.

Polina Tarantsova, 2^nd-year student of Software Engineering

I am doing research on Process Mining. The technologies developed by researchers in this sphere are widely used in businesses to improve economic efficiency andto analyze the behavior of users and systems.

Any process can be described as a sequence of events. In real life, we have to deal with huge data sets, which include hundreds of thousands events that are very hard to analyze. A process mining researcher works with an event log. Their task is to build a model that would describe the system from which the log has been retrieved in the most precise and compact way.

In 2018, my academic supervisor and I suggested an approach based on detecting sub-processes in a system, which allowed decreasing the size of the describing model by 65 times, without losing any behavioural patterns! As a result of the study, we’ve written a paper, which was published in a Springer journal and presented at AIST, an international conference. Today, I’m involved in the application of machine learning for retrieving models at the Laboratory of Process-Aware Information Systems (PAIS Lab). It is an interesting project with a huge ground for research, so I’m planning to further develop my work at the intersection of process mining and machine learning.

 

Artyom Gadetsky, 1^st-year student of the Master’s programme in Statistical Learning Theory

I’ve been doing research at Bayesian Methods Research Group headed byDmitry Vetrov for three years: two years as an undergraduate student, and now continuing as a Master’s student. My undergraduate thesis was an experimental proof of the fact that the contemporary neural networks retrieve much more information on data than was previously believed. A neural network not only understands the meaning of words depending on the context, but is able to formulate its understanding in human language. This opens new opportunities for AI interpretation. The results of this study have been presented at the world’s main conference on computer linguistics, ACL (rank A*). The technology offered in the paper is both practically applicable, for example in finance, and relevant as fundamental research.

I’m currently working in Structure Learning, when you have to build a cyclic oriented graph of relationship between variables. This may be applicable in many areas, where you need to be able to evaluate the cause-and-effect links instead of conditional distribution.

 

Roman Nesterov, doctoral student, Research Assistant at the Laboratory of Process-Aware Information Systems (PAIS Lab)

I am in my second year of the joint doctoral programme with the University of Milano-Bicocca(Università degli Studi di Milano-Bicocca, UNIMIB). I carry out research in synthesis of formal (executable) models of processes by information system logs (process discovery). This field is developing rapidly; in 2019, it was named an IT area of top global priority by Gartner.

Together with researchers from the group ‘Models of Concurrency, Computation and Communication’ at UNIMIB, we are working on methods for composition synthesis of multi-agent system models by their event logs. An important task for any approach that uses composition is providing ‘zero defects’ in the received models. For example, it is desirable that the model doesn’t allow deadlocks.

In 2018, we presented a general approach to composition synthesis at a PETRI NETS 2018 conference workshop. Our approach allows us to create a multi-agent system model by means of operated composition of specific agents’ behavior models, as well as inheriting the agents’ features. This means that it’s unnecessary to check the model of the whole system, if the agents’ models are zero-defect. In addition, the system model visually reflects the structure of agents’ interaction, which facilitates its application and analysis.

The application of this approach may require quite a deep theoretical knowledge, so we have also developed special composition patterns that are convenient for practical application. These patterns describe some typical ways of interaction between agents in multi-agent systems and are, in fact, ready-to-use solutions. We’ve used this developed library of patterns to conduct an in-depth experimental evaluation, the results of which confirm applicability of our approach. Today, we are actively working on clarifying and generalizing the results we’ve received.