Ilya Segalovich Scholarship Winners Interviews. Part 1

The winners of the Ilya Segalovich Scholarship were announced recently. We talked to them about the award, their studies and their projects.

Maria Marchenko
Second-year student of the Data Science and Business Analytics bachelor's programme

I am currently a second-year student in the Data Science and Business Analytics programme. Before that, I studied mathematics, computer science and engineering at the HSE Lyceum for two years. It was very interesting: there were many lectures in mathematics and computer science, each lyceum student had to write a thesis. It is there that I started working with big data. I really liked it and decided that I wanted to learn more about data science, so I went to Data Science and Business Analytics.

Now my studies are simple: it's easier with the subjects where you have to understand a concept and, based on that, get the results you need. At the end of the first year we had algebra, and I liked it so much that I decided to keep doing it. In July I started writing a term paper on applying Gröbner bases to geometry problems, I finished in the autumn and told Dmitry Trushin that I wanted to learn more about rings. He recommended some books to me and then suggested I write another term paper based on the material I had studied. I also took a special course on homological algebra in winter. I work as a front-end developer at a startup, although it can sometimes be hard to combine work and study.

When I heard about the Ilya Segalovich Scholarship nomination I got excited, but I thought I had little chance because there are a lot of people around who are also working on interesting topics. When I was named at the award ceremony, I was very surprised and happy. I didn't believe to the last minute that I was going to be a prizewinner. At the event, I remembered most of all the speech given by Elena Bunina. It was very interesting to learn about Ilya Segalovich, about what kind of person he was, and on the whole, her story was very sincere and heartfelt.

Yunging Pei
Third-year student of the Data Science and Business Analytics bachelor's programme

When I got the message from the study office at the beginning of March that I had been nominated for the Ilya Segalovich Scholarship, I was thrilled. I didn't expect to win at all, because there are so many cool and smart students in our department. 

Winning was a huge and unexpected surprise for me. My name was called last of the bachelors. When the name of the ninth laureate was called out, I was already one hundred per cent sure that I wouldn't get it. So when I heard it, I could not believe my ears at first, but then I was surrounded by joy and positive emotions for the whole evening. I am very grateful to the Faculty and my teachers, it is a great honour for me.

I really enjoyed the awards ceremony, it was not long and boring as such events often are, but instead very lively. I particularly remembered the story about Ilya Segalovich that Elena Bunina told us. Having listened to the story, I better understood the meaning of the scholarship and also began to feel great respect for Ilya. I hope I can follow in his footsteps and contribute to society in the future. 

I think that immersing yourself in your studies diligently is certainly important, but apart from attending classes and doing your homework, it's just as important to actively participate in projects and student activities such as teaching assistantships, mentoring and so on. The main thing is to constantly try to get better, be motivated and have faith in what you are doing.

Denis Rakitin
First-year student of the Math of Machine Learning master's programme

I learned about the Faculty of Computer Science in 2016, when I enrolled in the Applied Mathematics and Information Science programme. These four years were very cool, I got to know many interesting people and different directions in computer science. Most of all I was influenced by the probability theory course and electives by Dmitry Shabanov, Bayesian methods course by Dmitry Vetrov and working as a study assistant. 

Now I am continuing my studies in the Math of Machine Learning, a joint master's programme with Skoltech. Research activities are encouraged here, there is an emphasis on fundamental knowledge and fine theoretical courses. It's been almost a year since I started, and I'm loving it. Also, I am fully involved in the Bayes Group, this year I took part in teaching a course on neuro-Bayesian methods, and I am teaching a probability theory seminar to second-year students of Applied Mathematics and Information Science. The Faculty has had a profound impact on my life, and I am very happy to be part of it, now both a student and employee. 

Together with my  Bayes Group colleagues, we work on deep learning models with discrete latent variables. Such models are interesting in their own right, but also have great practical potential. For example, by adding a structural unit to a neural network, we give it a priori knowledge of what the data look like, which can improve the quality of the model, make it richer and more interpretable, and speed up the running time of some architectures. The problem with such models is that their discrete nature makes effective gradient estimation in the training phase very difficult. My colleagues and I are in the process of constructing and analysing gradient estimators as applied to various frameworks. There is quite a lot of scope in this topic, and the practical difficulties encountered only fuel the interest in trying to overcome them.

I was very pleased to be nominated for the Ilya Segalovich Scholarship. I often heard about it, saw friends and classmates being nominated, and thought that it would be cool to participate at some point. I can't say I expected or was sure of winning - you can't know in advance how things will go. Besides, we have plenty of talented kids in the Faculty. But at least I had my hopes up. 

The ceremony was great. Nice hall on Pokrovka Street, great atmosphere, familiar faces, pleasant and informative introduction and a lecture after the awarding, the ceremony itself, chatting and a cake after - it was a great evening! It was nice to see my good friends and colleagues among the awardees. 

I think the best advice for future laureates is to find an activity that you can really enjoy, that will keep your eyes sparkling, and you have to work hard. I think that almost any goal can be achieved if you are motivated and willing to put in the effort.

Ivan Safronov
Third-year student of the Applied Mathematics and Information Science bachelor's programme

I really enjoyed the ceremony, and I particularly remember the kind and interesting words about Ilya Segalovich. He is one of the people I aspire to. I had heard and read about him before, but this time I learned a great deal more. 

When I found out that I was nominated, I felt that my efforts were recognised and rewarded, which is certainly very motivating. Of course, I really wanted to win, but it was impossible to know in advance because all winners were worthy of winning.

My path to the Faculty was quite long. At first, I studied mathematics at St. Petersburg State University, but  I realized that I did not enjoy pure mathematics enough. I decided to transfer to the Faculty of Computer Science, to the Applied Mathematics and Computer Science programme. Now that two and a half years have passed, I feel that I made the right choice. Studying here is not easy at all: almost all the knowledge is new to me, so it's difficult but interesting to learn.

I feel like I've been rewarded for my academic and competition success. In addition to the Faculty, I study Big Data Infrastructure at Yandex School of Data Analysis. I actively participate in programming competitions, in particular in the ICPC World Championship for Students.

Speaking about the research, the topic of my term work is "Extraction of weighted finite automata from recurrent neural networks using tensor train decomposition". It lies at the intersection of linear algebra and machine learning. The work is concerned with the implementation and investigation of a new algorithm for extracting a weighted finite state machine from a recurrent neural network. This helps to interpret already trained complex neural networks that cannot be interpreted.

Korney Tomaschuk
Third-year student of the Applied Data Analysis master's programme

Since my second year, I have been working on a research project "Deep Learning Scenarios: Neurodifferential Equations and their Parameters" under the supervision of Vasily Gromov. In short, anyone who has trained neural networks is familiar with the problem of selecting the optimal hyperparameters for models - at the moment there is no effective algorithm for this. Recently a new breakthrough in deep learning has been introduced - neurodifferential equations that allow a neural network to be represented as a system of ordinary differential equations. Thus, the choice of hyperparameters for such a model can be formulated as a bifurcation theory problem and eventually an algorithm can be created to select the optimal hyperparameters of the network - this is what I do. 

At the awards ceremony, I did not notice the time flying. During the first part of the ceremony, I could not think about anything but the announcements, and afterwards, I was pleasantly shocked. Still, I would like to express my gratitude to Andrey Sebrant for an extremely interesting and informative lecture. When I heard my name announced I was overjoyed. It is extremely difficult to describe such strong emotions, so I will say this: I believe that for me it is the greatest achievement at the moment. 

The decision to enrol in the Faculty of Computer Science and get involved in mathematics was sudden for me as well as for my parents. It's worth starting with the fact that I went to school without a background in computer science and mathematics. Up until the end of school, I could not imagine that I would be a mathematician, even if it was applied mathematics, and that I would be doing research. At that time I was not fully aware of the direction in which I wanted to go further, so I, like most of my classmates, was planning to go abroad to study economics and mathematics. It is worth noting that the Moscow School of Economics where I studied, specialises in preparing students for entry into foreign universities. Because of this, I studied for twelve years instead of the usual eleven, and the last two years were aimed at preparing for IB international examinations. I even managed to enter Warwick University (UK). But at the last moment, I decided to abandon this idea and stay in Moscow to continue my studies at the Faculty. 

In the summer after the 11th grade, I decided to start learning to program for the interests’ sake. It all started with HSE University's Introduction to Python Programming by Coursera. I liked it so much that I began to dive into this field. This is how I learned about data science and machine learning and decided that I wanted to walk in this direction. I learned about the Faculty from my maths teacher. That year it was just announced that a new double-degree programme in Data Science and Business Analytics opened. Looking at the course curriculum, I knew straight away that I would apply exactly there. Studying is not very easy for me. But from the very first day here I knew that this was what I wanted to do.