Two HSE Researchers Receive Yandex ML Prize
Two researchers from the HSE Faculty of Computer Science—Aibek Alanov, Junior Research Fellow at the Centre of Deep Learning and Bayesian Methods, Research Fellow at AIRI; and Pavel Braslavski, Associate Professor, Senior Research Fellow at the Laboratory for Models and Methods of Computational Pragmatics—are among the winners of the 2023 Yandex ML Prize (formerly the Ilya Segalovich Award).
The award, established by Yandex in 2019, is aimed at supporting young researchers and the scientific community in the field of IT. Scientists who have presented the most promising and significant works in the field of speech recognition and synthesis, computer vision, information retrieval, natural language processing and generative models can apply for the award. This year, more than 200 applications from researchers from 11 countries were submitted. The award committee selected 11 laureates in 6 categories.
The laureates received prizes ranging from 500,000 to 1 million roubles depending on the category. In addition to the cash prize, Yandex provided the laureates with access to Yandex 360 services and issued a grant to use the Yandex Cloud platform, which can help researchers perform large calculations and process experimental data. This year, the size of the grant for cloud computing has been increased to 500,000 roubles.
The competition is open to bachelor’s, master’s and doctoral students studying at universities or research institutes in Russia, Azerbaijan, Armenia, Belarus, Kazakhstan, Kyrgyzstan, Moldova, Serbia, Tajikistan, Turkmenistan, and Uzbekistan. Researchers, academic supervisors and teachers who contribute to the development of machine learning can also participate in the contest.
At the HSE AI Research Centre, Aibek Alanov is working on the development and verification of conditional generation models to improve sound and image quality (the project is implemented with the support of MTS AI). He explores ways to adapt various modifications of diffusion probabilistic models to data generation tasks of various natures and internal structures.
For two years in a row, articles co-authored by Aibek were accepted at NeurIPS, the largest international conference on neural networks and artificial intelligence. Aibek co-authored the articles ‘Star-Shaped Denoising Diffusion Probabilistic Models’ in 2023 and ‘HyperDomainNet: Universal Domain Adaptation for Generative Adversarial Networks’ in 2022.
‘During my undergraduate and doctoral studies at the Faculty of Computer Science, I received the Ilya Segalovich scholarship twice, so I found out about the Yandex ML Prize immediately. I was very pleased to be so highly appreciated, and it gave me the motivation to work even harder. I felt that my work was important and valuable, and it gave me more confidence in myself and in my work.
The Yandex ML Prize is designed to support scientists engaged in fundamental research in the field of artificial intelligence. I think many people understand that without it, significant progress in the development of new technologies in this area is impossible. Therefore, it is especially important for the industry to support as many scientists as possible.
I think that in the future, artificial intelligence technologies will be an even larger part of our daily lives and radically change our way of life for the better. I don’t want just to consume these technologies, but to create and develop them. In my academic career, my main goal is to create my own research group that would conduct research at the international level and show results that are interesting and valuable for the entire field.’
Together with Ilya Segalovich, one of the creators of Yandex, Pavel Braslavski organised the first series of machine learning competitions in Russia. Pavel is also the organiser of the Russian Summer School in Information Retrieval (RuSSIR).
Pavel's article ‘You Told Me That Joke Twice: A Systematic Investigation of Transferability and Robustness of Humor Detection Models’ was accepted to the main programme of the EMNLP 2023 international conference on natural language processing. In this paper, the authors explore what data is needed to train humour recognition models, and what limitations these models have.
‘I want to thank Yandex for the award! A lot of things in my life started with working for this company, which I joined 18 years ago. Therefore, receiving the Yandex ML Prize seems to bring me back to the people I love. I also want to thank my family. Thanks for supporting science!’
Read about the Yandex Award 2022 here.