First Cohort Graduates from Master’s Programme in Statistical Learning Theory
The Master's Programme ‘Statistical Learning Theory’ was launched in 2017, and is run jointly with the Skolkovo Institute of Science and Technology (Skoltech). The programme trains future scientists to effectively carry out fundamental research and work on new challenging problems in statistical learning theory, one of the most promising fields of science. This field lies at the intersection of various disciplines of mathematics and computer science: mathematical statistics, machine learning, optimization, information theory, complexity theory, and others. From the early stages of preparing their theses, students collaborate in group research. Graduates of the programme receive a double degree—one from HSE University and one from Skoltech.
In 2019, the first cohort of MS students in 'Statistical Learning Theory' graduated from HSE, and seven of them graduated with distinction. Graduates Yury Kemaev and Maxim Kaledin sat down with HSE News Service to discuss the programme, give some advice to prospective students, and share their plans for the future.
What did you gain during your studies at the programme?
Yury: I applied to the programme after getting my bachelor’s in Computational Mathematics and Cybernetics from Moscow State University, where I studied in the Department of System Programming. While studying in the 'Statistical Learning Theory' programme at HSE, I gained knowledge that enables me to conduct research in the field of artificial intelligence.
Maxim: Before entering the programme, I had a little bit of an idea of what I wanted to do in the future, and here I made my final decision: I will do research in modern stochastics and multidimensional statistics. In many ways, the HSE lecturers, who we soon started working with at the HDI Lab, influenced my choice.
My colleagues always support me, we organize seminars and meetings together – this is how research works. This programme is a great way to become a part of real research
How does it feel to study at two universities at once?
Yury: It turned out to be very convenient, since you could receive support from both universities. Skoltech offered a good scholarship, and HSE University provided me with housing in the dorm. There are other good things. For example, HSE let me take interesting courses on Coursera and reimbursed me for the costs of participating in a conference. In my opinion, everything was organized perfectly, because it was possible to receive course credits from both universities, as well as from the Yandex School of Data Analysis. Other students and I took advantage of this opportunity. If you know that a certain course won’t be of any value to you, you can replace it with a different one. You can also get a group together and propose to replace one course with another, which is what we did with Neuro Bayesian Methods in our group. In my opinion, you should value this opportunity to manage your time and resources, especially during your master’s studies.
Maxim: It’s a challenge in terms of both bureaucracy and content. There were scheduling mix-ups sometimes, we had to do credit transfers and negotiate various formalities. It's nice that the administration always met us halfway and all these issues were resolved as well as possible.
Given that it’s a joint programme, there are twice as many requirements: you are literally a student at two different places
You need to attend the required courses, which are part of the curricula at both universities during the first year, and to pass exams. During the second year, you can choose your own courses, so you can make your schedule more flexible. In general, it’s not as difficult as it may seem. Some of the students even managed to have jobs.
HSE and Skoltech are fundamentally different universities with different views on education and programme curriculum. At HSE we had more mathematics courses related to probability and statistics—of particularly note were ‘Modern Stochastics’ (with Denis Belomestny and Alexey Naumov), and ‘High-Dimensional Statistics’ (with Quentin Paris). At Skoltech we had more applied courses. For example, I especially liked the courses ‘Numerical Linear Algebra’ and ‘Fast and Efficient Solvers’ with Ivan Oseledets. Some of the students preferred more practice-oriented work, while others, like me, became more engaged with theory. Some of us prefer Skoltech, and others prefer HSE. But we all learned a lot about both modern theory and its applications.
Programme graduates and Ivan Arzhantsev, Dean of the Faculty of Computer Science
© HSE University
What advice would you give to prospective students?
Yury: I would advise that they think about what they want to achieve in 2 years of master’s studies and whether they need it. Here is a thought experiment: here I am, a week after graduation. What can I do and what opportunities do I have now? Do I really want this? Why and what for? How will I achieve it during these two years? I didn’t have clear answers to all these questions at the time, but it helps to determine your general way, and this, in turn, allows you to set priorities for yourself.
Maxim: I would advise applicants to be more active, communicate with professors, attend research seminars, and search for conferences and schools on these topics. They should find a research team that shares their interests. As for the laboratories I know, I can say that they can do anything from object recognition and change detection in pictures to stochastics and financial mathematics.
I would say that the main advantage of this programme are the live research seminars. Everything is open; you just need to understand what you are interested in and study it. There will always be people who will be happy to work with you.
Can you tell us a little bit about the research groups?
I suppose the largest portion of our programme was made up of group work, and I think that is how it should be
There are other very strong groups both at Skoltech and HSE, their members will probably tell you more.
Maxim: Сourses in this programme are not as important as seminars and research groups. I remember that during the second year of studies I had two subjects in the first semester and one in the second. At the same time, I was spending a full day in the laboratory and was busy solving problems that were then partially included in my thesis. This is your main activity; in fact, it is essentially like an internship. There are several groups, and some of them were transformed into laboratories. A few that I know well are, for example, the HDI Lab where people do statistics, MCMC (Markov Chain Monte Carlo), and optimal transport. There is also the Centre of Deep Learning and Bayesian Methods with Dmitry Vetrov, where several of the students from my group worked.
At Skoltech there is a group led by Ivan Oseledets that works with multidimensional computational mathematics (both theoretical and practical issues) and applications of tensor methods to various engineering problems; they have many joint projects with the industry. At Skoltech there are still a few more groups in the applications of machine learning. Each group organizes research seminars and invites guest speakers to deliver lectures and mini-courses. I think anyone interested in statistics or data science would definitely find a team that shares his/her interests.
What about your future plans?
Yury: In September I joined the team of Google Deep Mind as Research Engineer, where I’ll be working on artificial intelligence technologies.
Maxim: My research supervisors Denis Belomestny and Eric Moulines and I are currently working out all the details for my cotutelle doctoral degree between the HSE Faculty of Computer Science and École Polytechnique. Right now, we are preparing all the necessary documents. In practice, this means that I will spend six months in Paris and six months in Moscow for three years, working on my dissertation and attending doctoral courses.
I will study reinforcement learning theory and statistics on manifolds – there will be enough work for many years. We want to understand how you can evaluate the reliability of RL algorithms and how you can accelerate learning (current algorithms require a lot of data, patience and tricks). In the field of manifolds, we are interested in how one can estimate distributions and sample nontrivial objects lying on manifolds (for example, covariance matrices of special structure that must be evaluated for Frequency Division Duplex in base stations).
My academic supervisors and the laboratory staff members (the HDI Lab at HSE and CMAP at Ecole Polytechnique) are experts in statistics and stochastics, and I am very proud to work with them. It’s hard to make plans for after doctoral school, but I hope that I will continue to work in science and collaborate with my colleagues from the HDI Lab.
How to apply
Applicants who wish to study in the joint programme between HSE and Skoltech must apply to both universities separately. If they so choose, applicants may apply to HSE only and sit in on classes free of charge at Skoltech (as non-degree seeking students), which allows students to study all courses offered by the SLT track.
Admissions to HSE’s programmes are now open. International students can apply online. To learn more about HSE University, its admission process, or life in Moscow, please visit International Admissions website, or contact the Education & Training Advisory Centre at: email@example.com, or via WhatsApp at: +7 (916) 311 8521.
Leading Research Fellow at the International Laboratory of Stochastic Algorithms and High-Dimensional Inference
Research Assistant at the International Laboratory of Stochastic Algorithms and High-Dimensional Inference
Academic Supervisor of the International Laboratory of Stochastic Algorithms and High-Dimensional Inference
Staff members of the HSE Faculty of Computer Science will present 12 of their works at the 37th Conference and Workshop on Neural Information Processing Systems (NeurIPS), one of the most significant events in the field of artificial intelligence and machine learning. This year it will be held on December 10–16 in New Orleans (USA).
On November 17-19, The HSE Faculty of Computer Science, SBER and cloud technology provider Cloud.ru organised HSE Sber ML Hack, a hackathon based around machine learning. More than 350 undergraduate and graduate students from 54 leading Russian universities took part in the competition.
Over three days, more than 300 conference participants attended workshops, seminars, sections and a poster session. During panel discussions, experts deliberated on the regulation of artificial intelligence (AI) technologies and considered collaborative initiatives between academic institutions and industry to advance AI development through megaprojects.
Machine Learning (ML) is a field of AI that examines methods and algorithms that enable computers to learn based on experience and data and without explicit programming. With its help, AI can analyse data, recall information, build forecasts, and give recommendations. Machine learning algorithms have applications in medicine, stock trading, robotics, drone control and other fields.
Based on the results of a project competition, two new laboratories are opening at HSE University’s Faculty of Computer Science. The Laboratory for Matrix and Tensor Methods in Machine Learning will be headed by Maxim Rakhuba, Associate Professor at the Big Data and Information Retrieval School. The Laboratory for Cloud and Mobile Technologies will be headed by Dmitry Alexandrov, Professor at the School of Software Engineering.
Joint Project of Scientists from HSE University and Surgut State University to Help Prevent Recurrent Heart Attacks and Strokes
One of the winning projects of a competition held by HSE University’s Mirror Laboratories last June focuses on the use of machine learning technologies to predict the outcomes of acute coronary syndrome. It is implemented by HSE University’s International Laboratory of Bioinformatics together with the Research and Educational Centre of the Medical Institute at Surgut State University. Maria Poptsova, Head of the International Laboratory of Bioinformatics and Associate Professor at HSE University’s Faculty of Computer Science, talks about how this joint project originated, how it will help patients, and how work to implement it will be organised.
On August 28–30, HSE University’s Faculty of Computer Science held the 4th Summer School on Machine Learning in Bioinformatics. This year, 670 people registered for the event, and over 300 visited in person. The programme included lectures and seminars on various spheres of bioinformatics: applied bioinformatics and the bioinformatics of DNA, RNA, and proteins; elementary genomics; modern methods of data analysis and molecular biology. The lectures were complemented by practical tasks aimed at different levels of knowledge.
Fall into ML 2023: HSE University’s Faculty of Computer Science Organises Conference on Machine Learning
On October 26–28, HSE University’s Institute of Artificial Intelligence and Digital Sciences at the Faculty of Computer Science will hold a conference titled Fall into ML 2023 withsupport from the HSE University AI Research Centre . The event is dedicated to promising directions of fundamental AI development.
HSE Researchers Examine Wellbeing of Russian Social Media Users and Rank Public Holidays by Popularity
Researchers of the HSE Graduate School of Business trained a machine-learning (ML) model to infer users' subjective wellbeing from social media posts. Having processed 10 million tweets, the researchers compiled a rating of holidays celebrated in Russia based on their popularity. The New Year tops the list, but Russian-speaking users of Twitter are also happy to celebrate Defender of the Fatherland Day, International Women's Day, and Halloween. The study findings have been published in PeerJ Computer Science.
A group of researchers from the HSE Faculty of Computer Science and the Sber AI Lab has increased the speed of gradient boosting, one of the most efficient machine learning algorithms. The proposed approach will make it possible to solve classification and regression problems faster. The results of the work were presented at the NeurIPS conference.