Getting the best of Skoltech and HSE with the Math of Machine Learning Olympiad
Aleksandr Rubashevskii | Winner of the Math of Machine Learning Olympiad, current PhD student at Skoltech
Meet Aleksandr Rubashevskii, a Skoltech PhD student and graduate of a Skoltech-HSE joint Master's program Math of Machine Learning. After his Bachelor's studies, Aleksandr was in the search of a high-class machine learning experience. He applied for the Olympiad and has never looked back. Enjoy his story!
a journey to Skoltech
I am from Barnaul, the capital of the Altai Region. Although I studied at a linguistic gymnasium, starting in middle school I got carried away with olympiad-level mathematics and physics. That's how I learned about MIPT (Moscow Institute of Physics and Technology).
Closer to the third year of my Bachelor's studies, I thought that I would like to delve into programming, and at that moment I learned about the specialization of Yandex and MIPT in data analysis and machine learning. I was fascinated by this, so I also took courses in programming from other departments, I even managed to connect the final thesis with data analysis. When I decided to pursue this as a career and searched for opportunities, I found that Skoltech's joint program in Math of Machine Learning was the perfect fit. So, I applied to compete in the program's Olympiad.
Closer to the third year of my Bachelor's studies, I thought that I would like to delve into programming, and at that moment I learned about the specialization of Yandex and MIPT in data analysis and machine learning. I was fascinated by this, so I also took courses in programming from other departments, I even managed to connect the final thesis with data analysis. When I decided to pursue this as a career and searched for opportunities, I found that Skoltech's joint program in Math of Machine Learning was the perfect fit. So, I applied to compete in the program's Olympiad.
Math of Machine Learning Olympiad
I learned about the Math of Machine Learning Olympiad (formerly known as Statistical Learning Theory Olympiad) from VKontakte. I was very attracted by the opportunity to enter Skoltech. I think this is a great chance to practice and feel the format of Skoltech's selection process since winners of the Olympiad get actual admission to the Master's program as a prize. Plus, the program itself was interesting to me — I was attracted by the emphasis on mathematical disciplines in education, and the opportunity to study at two universities at once and get 2 diplomas. So I decided to give it a try.
Although I was very worried, my impressions from the Olympiad are only positive. My background partially prepared me for the challenge, but still, I had to remember a lot from my first years of studies. I took those books on math analysis, linear algebra and probability theory and analyzed the theory, plus I looked at homework and notes on mathematical subjects. As far as English is concerned, I decided to pass the IELTS in advance to focus only on mathematics before the Olympiad. School English background, MIPT lessons and Skype lessons with a teacher from my city helped to prepare.
As a result of participating in the Olympiad, I became a student in the joint Math of Machine Learning Master's program of Skoltech and Higher School of Economics.
Although I was very worried, my impressions from the Olympiad are only positive. My background partially prepared me for the challenge, but still, I had to remember a lot from my first years of studies. I took those books on math analysis, linear algebra and probability theory and analyzed the theory, plus I looked at homework and notes on mathematical subjects. As far as English is concerned, I decided to pass the IELTS in advance to focus only on mathematics before the Olympiad. School English background, MIPT lessons and Skype lessons with a teacher from my city helped to prepare.
As a result of participating in the Olympiad, I became a student in the joint Math of Machine Learning Master's program of Skoltech and Higher School of Economics.
Math of Machine Learning Olympiad
Skoltech and Higher School of Economics invite students in their last year of math- and IT-related bachelor's studies to compete in solving advanced challenges in machine learning as part of the Math of Machine Learning Olympiad 2023.
Winners of the Olympiad will get a huge round of applause, prizes from the organizers and automatic enrollment into the joint Master's program of HSE and Skoltech "Math of Machine Learning". Winning the Olympiad is equivalent to passing the program's selection for admission. Learn more
Winners of the Olympiad will get a huge round of applause, prizes from the organizers and automatic enrollment into the joint Master's program of HSE and Skoltech "Math of Machine Learning". Winning the Olympiad is equivalent to passing the program's selection for admission. Learn more
impressions of MSc studies at Skoltech
I remember several things from Skoltech Master's program. Speaking about specific courses, I liked Ivan Oseledets' course on linear algebra and Evgeny Burnaev's course on machine learning. They had a huge emphasis on practice, so taking these courses helped a lot for the successful writing of a thesis and a deeper understanding of the subject area.
At Skoltech, I also remember working with my scientific advisor Dmitry Dylov. We developed a project on the detection and adjustment of the forearm vein mask in the near-infrared light domain. As a result, we even published a conference paper about this study. I think this experience pushed me to further my PhD.
Also, I have pleasant impressions from the summer industrial practice. I interned at Nvidia and worked on the detection of text blocks in screenshots of game images to create an automated testing bot. I gained valuable knowledge about imaging and deep learning in real-world tasks, as well as simply experienced the industry and understanding the internal processes of companies. This experience can be useful both in future work in the industry and in academic activities.
At Skoltech, I also remember working with my scientific advisor Dmitry Dylov. We developed a project on the detection and adjustment of the forearm vein mask in the near-infrared light domain. As a result, we even published a conference paper about this study. I think this experience pushed me to further my PhD.
Also, I have pleasant impressions from the summer industrial practice. I interned at Nvidia and worked on the detection of text blocks in screenshots of game images to create an automated testing bot. I gained valuable knowledge about imaging and deep learning in real-world tasks, as well as simply experienced the industry and understanding the internal processes of companies. This experience can be useful both in future work in the industry and in academic activities.
the benefits of the Skoltech-HSE joint program
Skoltech's joint program with the Higher School of Economics provided me with more choice of subjects, an opportunity to interact with HSE teachers, to write my thesis with them. Many of my classmates did this, and some even tried themselves in joint projects with teachers from both universities. One of the most helpful additional courses I took was Economics at HSE — it broadened my horizons beyond my subject area. HSE also has its own joint projects with foreign universities, such as exchange programs, for which you can also apply.
current PhD studies
I am now in Pavel Osinenko's group, where I study reinforcement learning with formal guarantees of safety and stability. Reinforcement learning is a subfield of machine learning that has been gaining in popularity lately. It has shown its strength by beating a man in chess, Go, Dota and other competitions.
Moreover, in this area, in most cases, there are no formal guarantees of success — everything is confirmed only experimentally. Theoretical works on this subject have begun to appear recently. Ideally, our plan is to develop a symbiosis of theoretical and experimental methods so that reinforcement learning can be fully utilized in the industrial tasks. We try to do this at the intersection of classical control theory, on which most of the current industrial autopilots are based, and reinforcement learning.
Moreover, in this area, in most cases, there are no formal guarantees of success — everything is confirmed only experimentally. Theoretical works on this subject have begun to appear recently. Ideally, our plan is to develop a symbiosis of theoretical and experimental methods so that reinforcement learning can be fully utilized in the industrial tasks. We try to do this at the intersection of classical control theory, on which most of the current industrial autopilots are based, and reinforcement learning.
takeaways and future plans
I really like that the study of artificial intelligence is now at a very high level and is supported around the world. Developments in this area can be done both in research universities and in laboratories of IT corporations. I have not yet decided exactly which of these I gravitate more towards, but I would definitely like to get international experience in research work. This is an opportunity to learn something new from colleagues from abroad and also an opportunity to live in other countries, to see what science looks like there.
The Math of Machine Learning Olympiad was a great step in my academic and professional career: not only did I gain invaluable experience and knowledge in the subject, but also had a chance to get the best opportunities from both universities — Skoltech and HSE.
The Math of Machine Learning Olympiad was a great step in my academic and professional career: not only did I gain invaluable experience and knowledge in the subject, but also had a chance to get the best opportunities from both universities — Skoltech and HSE.
JOIN THE OLYMPIAD
How to participate
Graduates and senior bachelor's students pursuing Bachelor's, Specialist's or Master's degree in Applied Mathematics and Information Science or related fields can participate in the Olympiad.
Winners of the Olympiad will get a huge round of applause, prizes from the organizers and automatic enrollment into the joint Master's program of HSE and Skoltech Math of Machine Learning. Winning the Olympiad is equivalent to passing the program's selection for admission.
Winners of the Olympiad will get a huge round of applause, prizes from the organizers and automatic enrollment into the joint Master's program of HSE and Skoltech Math of Machine Learning. Winning the Olympiad is equivalent to passing the program's selection for admission.
more stories? sure: