Faculty of Computer Science thrives and flourishes even during the pandemic. We have already told you how the Faculty switched to distance learning. However, the Faculty is a research centre as well as the educational one. It comprises twelve laboratories researching into data science, theoretical computer science, process-oriented information systems, bioinformatics, financial technologies, and other fields of study. We want to tell you about the progress of research at the Faculty of Computer Science during the first half of 2020.
This year, two new laboratories were established at the Faculty.
International Laboratory of Algebraic Topology and Its Applications was created in February. It is to continue and develop the work of the Laboratory of Applied Geometry and Topology created last year. The new position will allow the Laboratory to cooperate more actively with the international centres of algebraic topology. Anton Ayzenberg is the head of the Laboratory; its research supervisor is Professor Mikiya Masuda, of Osaka University, Japan.
Yandex Laboratory was launched in March with Artem Babenko as its head. The Laboratory researches on the problems of computer vision, natural language processing, machine translation, crowdsourcing, machine learning, speech technologies, information retrieval, and recommendations. Senior students will become the researchers in the Laboratory.
CEO of Yandex in Russia
Yandex and HSE University have been partners in many ways for more than ten years. Yandex Department and Laboratory at the Faculty of Computer Science show our commitment to both education and research. I am glad that we can further facilitate the development of computer science and education in collaboration with HSE University.
Before the quarantine, the Faculty was able to host several important events. In February, Math in Machine Learning 2020 school and conference was co-hosted by HSE University, Sirius University, and Yandex. The school included minicourses on deep learning (Lukasz Szpruch of The Alan Turing Institute, United Kingdom), reinforcement learning (Mikhail Valko, DeepMind), and optimal transport (Gabriel Peyré, CNRS, France).
The annual Moscow International School of Physics was organised in March by the Laboratory for Methods for Big Data Analysis (LAMBDA), Lebedev Physical Institute of the RAS, and Skoltech. Apart from lectures on physics, cosmology, and machine learning, the Young Scientist Forum was organised to present the results of new theoretical and experimental researches.
Among other events, second research workshop New Level of Visualisation – New Level of Analytics and fourth international workshop Formal Concept Analysis for Knowledge Discovery are worth mentioning.
The Faculty of Computer Science Colloquium and laboratories’ seminars continue to work as well.
The number of articles published by the Faculty’s researchers in Scopus and Web of Science journals demonstrates high research activity too. More than one hundred papers were published by the staff of the Faculty between January and June.
At the Faculty of Computer Science, one of the most important collaboration is that of LAMBDA and CERN, particularly, with LHCb experiment of the Large Hadron Collider. The report was published in January on the detection of excited omega baryons. The particles, heretofore undetected, were traced thanks to the LAMBDA’s algorithm.
Our researchers are highly valued at home, as well as abroad. In April, the Russian Science Foundation awarded grants to support the researches of Professor Nikolay Vereshchagin and Professor Alexander Shapoval.
Presentations and Talks
Over the past half-year, the researchers of the Faculty have given talks and presentations at several important international conferences. At LISC conference Professor Michael Vyalyi of HSE University and Dmitry Chistyakov of the University of Warwick, UK presented the article Re-pairing brackets.
Researchers of the Centre of Deep Learning and Bayesian Methods Artyom Gadetsky, Kirill Struminsky, Professor Dmitry Vetrov, Novi Quadrianto (HSE University, University of Sussex, UK) and Christopher Robinson of the University of Sussex gave the presentation on Low-Variance Black-Box Gradient Estimates for the Plackett-Luce Distribution at AAAI-20.
The members of the Bayes Group have also prepared several research works
- Pitfalls of In-Domain Uncertainty Estimation and Ensembling in Deep Learning (Arsenii Ashukha, Alexander Lyzhov, Dmitry Molchanov, Dmitry Vetrov) for ICLR 2020
- Deterministic Decoding for Discrete Data in Variational Autoencoders (Daniil Polykovskiy, Dmitry Vetrov) for AISTATS 2020
- Greedy Policy Search: A Simple Baseline for Learnable Test-Time Augmentation (Dmitry Molchanov, Alexander Lyzhov, Yuliya Molchanova, Arsenii Ashukha, Dmitry Vetrov) for UAI 2020
- Controlling Overestimation Bias with Truncated Mixture of Continuous Distributional Quantile Critics (Arsenii Kuznetsov, Pavel Shvechikov, Alexander Grishin, Dmitry Vetrov) and Involutive MCMC: a Unifying Framework (Kirill Neklyudov, Dmitry Vetrov, MaxWelling of the University of Amsterdam, Netherlands, Evgenii Egorov of Skoltech) for ICML 2020
- On Power Laws in Deep Ensembles (Ekaterina Lobacheva, Nadezhda Chirkova, Maxim Kodryan, Dmitry Vetrov) for Uncertainty and Robustness in Deep Learning Workshop, ICML 2020.
At the COLT 2020, the researchers of the International Laboratory of Stochastic Algorithms and High-Dimensional Inference Maksim Kaledin, Eric Moulines, Alexey Naumov, and Vladislav Tadic of the University of Bristol, UK and Hoi-To Wai of the Chinese University of Hong Kong presented the paper Finite Time Analysis of Linear Two-timescale Stochastic Approximation with Markovian Noise.
In February, the head of the Laboratory of Bioinformatics Maria Poptsova gave the talk Non-coding RNA in Heart Diseases: the Perspectives of Diagnostics and Therapy at the first international congress Genetics and Heart, which she co-chaired. In March, Arina Nostaeva, the research intern of the Laboratory, gave the poster presentation Patterns of promoter quadruplexes associated with different epigenetic marks at the Applied Bioinformatics in Life Sciences conference in Leuven, Belgium.
In February, LAMBDA researchers participated in the OpenTalks.AI conference. The head of LAMBDA Andrey Ustyuzhanin gave the talk Predictive analytics - state of the art and what was important in 2019. Senior research fellow Denis Derkach presented the tutorial Generative-Adversarial Networks: first steps in image generation.
Senior research fellow of the International Laboratory for Intelligent Systems and Structural Analysis Andrey Neznanov presented at Data Quality 2020 conference and fifth international conference Digital Media for the Future. He also recorded a lecture on highly immersive computer games.
We congratulate Tatyana Makhalova who has defended her thesis and received the degree of the Candidate of Science.
We are also happy to list those doctoral students who have had their preliminary defences – Pavel Sulimov, Dmitry Altukhov, Anna Sokolova, Vladislav Shakhuro, Sergey Shershakov, Maksim Borisyak, Nikita Kazeev, Alexander Turin, and Kirill Nekludov. All defences were conducted online.