Faculty's Research Digest 2020 (part 2)

It is a custom to tally up the year's results in December. We would like to tell you the results of the scientific life at the Faculty of Computer Science in 2020. You can read about the first half of the year in our previous digest. While this was not a year on which we shall look back with undiluted pleasure, but we still managed to achieve important results, adapted to changing circumstances, and made our best.

New departments and programmes

One of the highlights of this autumn was the creation of the Faculty's 1C Department, the online presentation of which was held on September 18. The Faculty has a long-standing record of cooperation with 1C, the company's employees teach at the Faculty, and every year 1C:Scholarship is awarded to students for their success in project activities.

Boris Nuraliev
1C CEO, 1C Department Head

Faculty of Computer Science is one of Russia's best training centre for software developers. We are happy to open our department within the School of Software Engineering and we will do our best to make our courses and disciplines interesting and useful for students... We already offer and will continue to offer project and thesis topics to the students of the Faculty, we will also continue to offer summer internships. To develop and support the project activities we have established 1C:Scholarship, and I would like to take this opportunity to congratulate the first winners. We look forward to fruitful cooperation and are optimistic about the future.

In addition to the new department, there is also a new Mirror Laboratory at the Faculty. The project 'Intelligent Data Analysis in Medical Information Systems' was created jointly by the International Laboratory of Intelligent Systems and Structural Analysis of HSE University and the Research and Education Centre of the Polytechnic Institute of Surgut State University. This project won the Mirror Lab contest, which took place from 15 April to 15 June 2020. The laboratory will research, among other things, mathematical modelling of the movement of fluids in the organs of the biological organism.

Bioinformatics Laboratory won the HSE University competition to create an international laboratory. The project 'Computational Genetics of Alternative DNA Structures' will be led by Alan Herbert (InsideOutBio, USA and University of Auckland, New Zealand) and Maria Poptsova, and the lab's status will change to 'international'.

Another project of the Bioinformatics Laboratory, the Consortium 'Genetics of cardiovascular diseases', was created jointly with KM-Alliance Association, Central State Medical Academy and Central Clinical Hospital with Polyclinic of the Administration of the President of Russia, Russian Medical Academy for Continuing Professional Education, Pirogov Russian National Research Medical University, Emanuel Institute of Biochemical Physics, Surgut State University, and Northern State Medical University.

Finally, the university has launched a joint master's-PhD programme that allows first-year master's students who wish to engage in research to begin their research work in advance and prepare for doctoral studies. This year, three students supervised by the Faculty's staff were enrolled in the Master's-PhD programme.

Conferences

Despite the impossibility of holding events in the usual format, the Faculty regularly organises conferences, workshops and schools online.

The Fourth International Conference on Computer Simulations in Physics and beyond (CSP-2020) was held on 12-16 October. It was organised by HSE University in cooperation with the Scientific Centre of the Russian Academy of Sciences in Chernogolovka and the Landau Institute of Theoretical Physics of the Russian Academy of Sciences. Ivan Arzhantsev, Dean of the Faculty, and Igor Agamirzyan, Professor of the Faculty, became members of the Conference Advisory Board; Andrey Ustyuzhanin, Head of the Laboratory of Methods for Big Data Analysis (LAMBDA), was a member of the Programme Committee; he also co-chaired the section "Algorithms, methods and tools with scalability and extended parallel modelling properties".

Talks

• Online detection of failures generated by storage simulator (Kenenbek Arzymatov, Mikhail Guschin, Andrey Sapronov, Vladislav Belavin, Leonid Gremyachikh, Maksim Karpov and Andrey Ustyuzhanin)

• Galaxies Clusters Reconstruction with Variational Autoencoder (from the Faculty: Ekaterina Trofimova, Alena Zarodnyuk, Dmitrii Gradoboev, Madina Atymkhanova, Andrey Ustyuzhanin and Alexey Solovyov)

• Fast simulation of the LHCb electromagnetic calorimeter response using VAEs and GAN (from the Faculty: Nikita Jain, Ivan Knunyants, George Kostenkov, Fedor Sergeev, Roman Bryanskiy and Ekaterina Trofimova)

• ML in Calorimeter optimization (Alexey Boldyrev, Denis Derkach, Fedor Ratnikov and Andrey Shevelev)

The ninth international conference Analysis of Images, Social Networks and Texts (AIST 2020), organised by HSE University and Skoltech, was held on October 15-16. Among the organisers were included Dmitry Ignatov, head of the Laboratory for Models and Methods of Computational Pragmatics, and Ekaterina Artemova of this Laboratory; Sergey Kuznetsov, head of the School of Data Analysis and Artificial Intelligence, and Ilya Makarov of this School; Irina Lomazova, head of the Laboratory of Process-Oriented Information Systems (POIS Lab), and Wil van der Aalst, honorary academic supervisor of the Laboratory.

Talks

• Human Action Recognition for Boxing Training Simulator (Anton Broilovskiy and Ilya Makarov)

• Object-Attribute Biclustering for Elimination of Missing Genotypes in Ischemic Stroke Data (from the Faculty: Dmitry Ignatov, Stefan Nikolic and Makhmud Shaban)

• Semi-Automatic Manga Colorization using Conditional Adversarial Networks (Maksim Golyadkin and Ilya Makarov)

• Automated Image and Video Quality Assessment for Computational Video Editing (Konstantin Lomotin and Ilya Makarov)

• ELMo and BERT in semantic change detection for Russian (from the Faculty: Yuliya Trofimova, Julia Rodina, Andrey Kutuzov and Ekaterina Artemova)

• Generating Sport Summaries: A Case Study for Russian (from the Faculty: Elena Tutubalina)

• DaNetQA: a yes/no Question Answering Dataset for the Russian Language (Ekaterina Artemova, Taisiya Glushkova, Alexey Machnev and Dmitry Ignatov)

• BERT for Sequence-to-Sequence Multi-Label Text Classification (from the Faculty: Ramil Yarullin)

• Detecting Automatically Managed Accounts in Online Social Networks: Graph Embedding Approach (Ilia Karpov and Ekaterina Glazkova)

• Federated Learning in Named Entity Recognition (Efim Luboshnikov and Ilya Makarov)

• Semantic Recommendation System for Bilingual Corpus of Academic Papers (from the Faculty: Irina Nikishina)

• Prediction of News Popularity via Keywords Extraction and Trends Tracking (from the Faculty: Alexander Pugachev and Ilya Makarov)

• Linking Friends in Social Networks using HashTag Attributes (Olga Gerasimova and Viktoriia Syomochkina)

• Community Detection Based on the Nodes Role in a Network: the Telegram Platform Case (Kseniia Tikhomirova and Ilya Makarov)

• How Minimal Degree of the Social Graph Affects Efficiency of an Organization (from the Faculty: Ilya Samonenko and Tamara Voznesenskaya)

• Data and Reference Semantic-Based Simulator of DB-nets with the Use of Renew Tool (Sergey Shershakov and Anton Rigin)

• Checking Conformance between Colored Petri Nets and Event Logs (Julio Carrasquel, Khalil Mecheraoui and Irina Lomazova). The Best Paper Award in Process Mining track.

The international conference Modeling and Analysis of Complex Systems and Processes (MACSPro-2020) was held on October 22-24. Among its organisers were Sasha Shapoval, head of the Laboratory of Complex Systems Modeling and Control, member of the Laboratory Victor Popov, Irina Lomazova and Sergey Shershakov.

Talks

• Compositional Conformance Checking of Nested Petri Nets and Event Logs of Multi-Agent Systems (Julio Carrasquel, Khalil Mecheraoui and Irina Lomazova)

• Content Based Video Retrieval System for Distorted Video Queries (Boris Tseytlin, Ilya Makarov)

• Software System Behavior Can Be Analyzed with Visual Analytics (Yaroslav Kotylev, Alexey Mitsyuk)

• Deep learning approach to recognize genome functional elements using diverse genomic data (Nazar Beknazarov, Seungmin Jin, Maria Poptsova)

• Educational Game Analysis Using Intention and Process Mining (Konstantin Nikitin)

• Real-Time Vehicle Type Detection and Counting from Road Camera Video (Denis Zuenko, Ilya Makarov)

• Model of General Equilibrium in Multisector Economy with Monopolistic Competition and Hypergeometric Utilities (from the Faculty: Sasha Shapoval)

•  Prediction after a horizon of predictability (Vasilii Gromov)

• Named Entity Recognition from Chernobyl Documentaries (from the Faculty: Ilya Makarov)

On November 18-20, the Faculty hosted a conference on machine learning, where staff, researchers and students presented papers in Research and Applied Machine Learning tracks.

Talks

​​​​​​

• Low-Variance Black-Box Gradient Estimates for the Plackett-Luce Distribution (Kirill Struminsky)

• On Power Laws in Deep Ensembles (Ekaterina Lobacheva)

• Machine learning as a tool quantifying the role of large sunspot groups in solar activity (Sasha Shapoval)

• Barycenters in metric spaces with applications (Quentin Paris)

Other talks were given by Maxim Kaledin, Alexey Naumov, Anton Osokin, Max Ryabinin, Maxim Rakhuba, Nikita Puchkin, Maria Poptsova, Ekaterina Artemova, Darya Maximova, Nadezhda Chirkova, Sergey Samsonov, Denis Derkach, Ksenia Volkova in Russian.

The international conference Topology and Geometry of Group Actions was held on November 18-22. It was organised by the International Laboratory of Algebraic Topology and its Applications of the Faculty of Computer Science and the Laboratory for Algebraic Groups of Transformations of the Faculty of Mathematics, HSE University.

Talks

• Invariants of the cohomology rings of the permutohedral varieties (Mikiya Masuda, academic supervisor of the Laboratory of Algebraic Topology)

• Substitution complexes and nonrealizability example (Semyon Abramyan)

• Topology of orbit spaces of complexity one torus actions in nongeneral position (Vladislav Cherepanov)

• Polyhedral products and commutator subgroups of right-angled Artin and Coxeter groups (Yakov Verevkinн)

• On polyhedral products over minimally non-Golod complexes (Ivan Limonchenko, member of the Laboratory of Algebraic Topology)

• B-root subgroups on affine spherical varieties (Roman Avdeev)

Schools

On July 16-30, Sixth Machine Learning in High Energy Physics Summer School (MLHEP) was held by the Laboratory of Methods for Big Data Analysis (LAMBDA) together with the Yandex School of Data Analytics and the Laboratory of High Energy Physics at the École Polytechnique Fédérale de Lausanne (EPFL). For the first time in six years, the school took place online, enabling it to host 114 participants from 23 countries and 73 universities - twice as many as usual.

On August 24-28, Summer School on Machine Learning in Bioinformatics was organised by the Laboratory of Bioinformatics for the first time. The school was attended by 126 participants from 22 countries and more than 60 universities. Among the speakers were Maria Poptsova, head of the Bioinformatics Laboratory, and Mikhail Gelfand, academic supervisor of Data Analysis in Biology and Medicine master's programme.

On September 11-13, the International Laboratory of Algebraic Topology and its Applications held an online school on Topology and Geometry Applications – 2020 with more than 300 participants from 17 countries. The school featured talks by Mikiya Masuda, academic supervisor of the Laboratory, Anton Aysenberg, head of the Laboratory, deputy head Vsevolod Chernyshev, and member of the Laboratory Fyodor Pavutnitsky.

On November 24-27, Second HSE-Yandex autumn school on generative models was held by LAMBDA, Centre for Deep Learning and Bayesian Methods and International Laboratory of Stochastic Algorithms and High-Dimensional Inference (HDI Lab) in cooperation with Yandex. The teachers of the school were Dmitry Vetrov, head of the Centre for Deep Learning and Bayesian Methods, deputy head Ekaterina Lobacheva, Centre's Nadezhda Chirkova, LAMBDA's Denis Derkach, and Quentin Paris from HDI Lab.

Seminars

The seminars of PAIS Lab, Laboratory for Models and Methods of Computational Pragmatics, Laboratory for Intelligent Systems and Structural Analysis, Mirror Lab, Laboratory of Complex Systems Modeling and Control, International Laboratory of Algebraic Topology and its Applications, and the Seminar on Machine Learning Methods in Bioinformatics continue their regular work.

In September, meetings of the inter-campus seminar on computer science began. At the seminar, staff from all campuses of HSE University talk about their research, discuss approaches to problem-solving and the application of developed methods in related areas. Faculty's researchers had already made three presentations at the inter-campus seminar.

On August 29, European Conference on Artificial Intelligence (ECAI 2020) hosted the eighth workshop "What Formal Concept Analysis can do for Artificial Intelligence?" (FCA4AI). Sergey Kuznetsov, head of the Data Analysis and Artificial Intelligence Department and head of the International Laboratory for Intelligent Systems and Structural Analysis co-chaired the programme committee of the workshop.

On September 18-19, as part of the Mirror Lab work, the seminar "Mathematical Modelling of Forward and Inverse Problems in Phlebology" was organized jointly by ISSA Lab and Surgut State University. Head of ISSA Lab Sergey Kuznetsov and Laboratory staff members Vasilii Gromov and Alexey Neznanov spoke at the seminar.

The eleventh international seminar Program Semantics, Specification and Verification: Theory and Applications (PSSV-2020) organized under the aegis of PAIS Lab took place on November 3-4.  Irina Lomazova, head of the Laboratory, was a member of the Programme Committee and organised the workshop together with Laboratory staff members Vladimir Zakharov and Roman Nesterov.

On November 4, Pierre Jacob (Department of Statistics, Harvard University) gave a presentation on Unbiased Markov chain Monte Carlo methods with couplings at the HDI Lab workshop.

On November 10, head of LAMBDA Andrey Ustyuzhanin together with LAMBDA's Denis Derkach and Artem Ryzhikov held a laboratory "Introduction to Generative Models" as part of the Archipelago 20.35 Intensive.

The work of the Faculty's colloquium, IT lectures and doctoral school seminar also continues.

Talks

Ilya Makarov, senior lecturer of the School of Data Analysis and Artificial Intelligence, presented a paper "Real-Time 3D Model Reconstruction and Mapping for Fashion" at the 43rd Telecommunications and Signal Processing International Conference (TSP), held on July 7-9. The presentation was based on joint research with Daniil Chernyshev, a graduate of Applied Mathematics and Computer Science bachelor's programme.

At the 37th International Conference on Machine Learning (ICML), held on July 12-18, Michal Rosenwald, research intern of the Bioinformatics Laboratory, presented a poster "Prediction of 3D Chromatin Structure with Machine Learning". Co-authors of the paper were Ekaterina Khrameeva (Skoltech), Grigory Sapunov (Intento) and Mikhail Gelfand. At the same conference members of the Bayesian Methods Research Group presented the papers  "Controlling Overestimation Bias with Truncated Mixture of Continuous Distributional Quantile Critics" and "Involutive MCMC: a Unifying Framework" described in our previous digest.

On 17 October, LAMBDA's Denis Derkach gave a lecture at the NAUKA 0+ Science Festival on the application of machine learning in particle physics.

On November 17, at the 30th annual symposium of the American Medical Informatics Association (AMIA 2020) Ella Tyuryumina, an associate researcher at ISSA Lab and senior lecturer in the School of Data Analysis and Artificial Intelligence, gave a presentation "A Mathematical Model to Predict the Diagnostic Periods for the Secondary Distant Metastases in Patients With ER/PR/HER2/Ki-67 Subtypes of Breast Cancer". The report was co-authored by ISSA Lab's Alexei Neznanov and Yakov Tyuryumin.

At the 63rd MIPT All-Russian Scientific Conference, HSE researchers presented three papers. LAMBDA's Andrey Ustyuzhanin and Oleg Serikov of the Laboratory of Models and Methods of Computational Pragmatics presented in Russian, while LAMBDA student Alexander Levin gave a presentation "Natural Language to Machine Learning in English on November 24.

At the 34th International Conference on Neural Information Processing Systems (NeurIPS 2020) on December 6-12, three papers of the Faculty's researchers were featured:

• Black-Box Optimization with Local Generative Surrogates by Vladislav Belavin and Andrey Ustyuzhanin of LAMBDA
• On Power Laws in Deep Ensembles by Ekaterina Lobacheva, Nadezhda Chirkova, Maxim Kodryan, and Dmitry Vetrov of HSE-Samsung Lab
• Towards Crowdsourced Training of Large Neural Networks using Decentralized Mixture-of-Experts by Max Ryabinin of Yandex Lab.

Achievements

In August, Dmitry Vetrov, head of the Centre for Deep Learning and Bayesian Methods, was elected to the European Lab for Learning and Intelligent Systems (ELLIS).

Russian Foundation for Basic Research supported two projects at the Faculty of Computer Science, "Ontological Access to Data Using the Disjunctive Axiom" by Olga Gerasimova (supervised by Mikhail Zakharyaschev), and "Development of Natural Language Processing Methods for Anomaly Detection in Big Data Storage Systems Event Logs" by Maxim Karpov (supervised by Andrey Ustyuzhanin).

In October, an article by the staff of the Centre for Deep Learning and Bayesian Methods won the bronze award at the Samsung Best Paper Award 2020 competition. The article "Controlling Overestimation Bias with Truncated Mixture of Continuous Distributional Quantile Critics" by Arseny Kuznetsov, Pavel Shvechikov, Alexander Grishin, and Dmitry Vetrov was the only paper from Russia to reach the final of the competition.

HSE University improved its position in the US News Best Global Universities rankings, taking 82nd place in mathematics, rising 23 points from last year and entering the top 100. In the institutional ranking, HSE University was ranked 547th, improving by 27 positions, and became the seventh in the list of Russian universities.

In December, LAMBDA researcher Mikhail Guschin became coordinator of the working group for machine learning and statistics in the LHCb Large Hadron Collider experiment at CERN.

Theses defences

Candidate's theses were defended at the Faculty by Maxim Borisyak, Nikita Kazeev, Anna Sokolova, Alexander Tyurin and Kirill Neklyudov.

On December 28, the first defense of a doctoral thesis at the Faculty of Computer Science will take place. The thesis is entitled "Numerical methods for optimising high-dimensional problems: Imprecise oracle and straightforward binary analysis"; it is prepared by Pavel Dvurechensky, senior researcher at the HDI Lab. The research supervisor of the thesis is Alexander Gasnikov, HDI Lab's senior researcher.