Research Digest 2021

Conferences	Faculty's events	New departments and projects

 

Conferences

AAAI

35th AAAI Conference on Artificial Intelligence (February 2-9)

• Improving Continuous-time Conflict Based Search (A. Andreychuk, Konstantin Yakovlev, E. Boyarski, R. Stern)
• Revisiting Mahalanobis Distance for Transformer-Based Out-of-Domain Detection (A. Podolskiy, D. Lipin, A. Bout, Ekaterina Artemova, I. Piontkovskaya)

AISTATS

24th International Conference on Artificial Intelligence and Statistics (April 13–15)

• Improved Complexity Bounds in Wasserstein Barycenter Problem (D. Dvinskikh, Daniil Tiapkin)

NAACL

Annual Conference of the North American Chapter of the Association for Computational Linguistics (June 6-11)

• A Simple Approach for Handling Out-of-Vocabulary Identifiers in Deep Learning for Source Code (Nadezhda Chirkova and Sergey Troshin)
• On the Embeddings of Variables in Recurrent Neural Networks for Source Code (Nadezhda Chirkova)

ICML

38th International Conference on Machine Learning (July 18-24)

• Newton Method over Networks is Fast up to the Statistical Precision (A. Daneshmand, G. Scutari, Pavel Dvurechensky, Alexander Gasnikov)
• ADOM: Accelerated Decentralized Optimization Method for Time-Varying Networks (D. Kovalev, E. Shulgin, P. Richtarik, Alexander Rogozin, Alexander Gasnikov)
• On a Combination of Alternating Minimization and Nesterov’s Momentum (S. Guminov, Pavel Dvurechensky, Nazarii Tupitsa, Alexander Gasnikov)

ACM SIGMOD/PODS

International Conference on Management of Data (June 20-25)

• Convergence of Array DBMS and Cellular Automata: A Road Traffic Simulation Case (Р. Антонио Родригес Залепинос)
• Deciding Boundedness of Monadic Sirups (Stanislav Kikot, A. Kurucz, Vladimir Podolskii, Mikhail Zakharyaschev)

ISSAC

46th International Symposium on Symbolic and Algebraic Computation (July 18-23)

• Wilf classes of non-symmetric operads (Andrey Cherkasov, Dmitry Piontkovsky)

SIGIR

44th ACM SIGIR Conference on Research and Development in Information Retrieval (July 11-15)

• A Systematic Evaluation of Transfer Learning and Pseudo-labeling with BERT-based Ranking Models (Iurii Mokrii, L. Boytsov, Pavel Braslavski)

EMNLP

Conference on Empirical Methods in Natural Language Processing (November 7-11)

• Artificial Text Detection via Examining the Topology of Attention Maps (L. Kushnareva, D. Cherniavskii, V. Mikhailov, Ekaterina Artemova, S. Barannikov, A. Bernstein, I. Piontkovskaya, D. Piontkovski, E. Burnaev)
• NB-MLM: Efficient Domain Adaptation of Masked Language Models for Sentiment Analysis (Nikolay Arefyev, D. Kharchev, A. Shelmanov)
• SPARQLing Database Queries from Intermediate Question Decompositions (Irina Saparina, Anton Osokin)
• Multi-Sentence Resampling: A Simple Approach to Alleviate Dataset Length Bias and Beam-Search Degradation (I. Provilkov, Andrey Malinin)
• Uncertainty Measures in Neural Belief Tracking and the Effects on Dialogue Policy Performance (C. van Niekerk, Andrey Malinin, Ch. Geishauser, M. Heck, H. Lin, N. Lubis, S. Feng, M. Gašić)

NeurIPS 2021

35th Conference on Neural Information Processing Systems (December 6-14)

• On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay (Ekaterina Lobacheva, Maxim Kodryan, Nadezhda Chirkova, Andrey Malinin, Dmitry Vetrov)
• Leveraging Recursive Gumbel-Max Trick for Approximate Inference in Combinatorial Spaces (Kirill Struminsky, Artyom Gadetsky, Denis Rakitin, D. Karpushkin, Dmitry Vetrov)
• Distributed Saddle-Point Problems Under Similarity (Aleksandr Beznosikov, G. Scutari, Alexander Rogozin, Alexander Gasnikov)
• Lower Bounds and Optimal Algorithms for Smooth and Strongly Convex Decentralized Optimization Over Time-Varying Networks (D. Kovalev, E. Gasanov, P. Richtárik, Alexander Gasnikov)
• Moshpit SGD: Communication-Efficient Decentralized Training on Heterogeneous Unreliable Devices (Max Ryabinin, Eduard Gorbunov, Vsevolod Plokhotnyuk, G. Pekhimenko)
• Tight High Probability Bounds for Linear Stochastic Approximation with Fixed Stepsize (A. Durmus, Eric Moulines, Alexey Naumov, Sergey Samsonov, K. Scaman, H. Wai)
• Distributed Deep Learning in Open Collaborations (Michael Diskin, A. Bukhtiyarov, Max Ryabinin, L. Saulnier, Q. Lhoest, Anton Sinitsin, Dmitry Popov, Dmitry Pyrkin, Maxim Kashirin, Alexander Borzunov, A. Villanova del Moral, D. Mazur, I. Kobelev, Y. Jernite, T. Wolf, G. Pekhimenko)
• Scaling Ensemble Distribution Distillation to Many Classes with Proxy Targets (Max Ryabinin, Andrey Malinin, M. Gales)
• Revisiting Deep Learning Models for Tabular Data (Y. Gorishniy, Ivan Rubachev, V. Khrulkov, Artem Babenko)
• Good Classification Measures and How to Find Them (M. Gösgens, A. Zhiyanov, A. Tikhonov, Liudmila Prokhorenkova)
• Overlapping Spaces for Compact Graph Representations (K. Shevkunov, Liudmila Prokhorenkova)
• Federated-EM with heterogeneity mitigation and variance reduction (A. Dieuleveut, G. Fort, Eric Moulines, G. Robin)

Look up selected publications by the Faculty's staff and students here.

A student of the Faculty Andrey Milko went to the UN Climate Change Conference of the Parties (COP26) in Glasgow on October 30 — November 12. He demonstrated a model of a city of the future that employed augmented reality features to demonstrate the nuclear power plant and its operation.

Faculty's events

From February to April, the Laboratory of Process-Aware Information Systems (PAIS Lab) ran a series of Wednesday Nights on Petri Nets and their Extensions workshops on process extraction and analysis.

Playlist

IDAO 2021 International Data Analysis Olympiad took place from March to April. It was organised by the Faculty in cooperation with Yandex; with Otkritie Bank as Platinum Partner. The first task was concerned with the search for dark matter; the second — provided by Otkritie Bank — was on personalised consumer loan offers.

A, B and Z: The Structure, Function and Genetics of Z-DNA and Z-RNA meeting was held on May 19, 2021. The first of its kind, this international conference gathered more than 200 scientists from 24 countries. The meeting was organised by the Faculty, Molecules (an open-access journal by MDPI), RNA Society, International Journal of Molecular Sciences, and InsideOutBio. Maria Poptsova and Alan Herbert were members of the program and organizing committee.

Full programme

On June 29, 2021, Cardiogenetics and Bioinformatics - Hypertrophic Cardiomyopathy School was held by the Genetics of Cardiovascular Diseases consortium, of which the International Laboratory of Bioinformatics is the member.

On July 15-30, the Seventh International School on Machine Learning in High Energy Physics (MLHEP) organised by the Laboratory of Methods for Big Data Analysis (LAMBDA), Yandex School of Data Analysis and EPFL (Switzerland) took place. This year's school attracted 121 participants, including representatives of 70 universities and research centres from Russia, Europe, and the USA.

Lectures and workshops

The Moscow Conference on Computational Molecular Biology (MCCMB) took place on July 30 — August 2 at Skoltech (Moscow, Russia). It was co-organised by the International Laboratory of Bioinformatics; Prof. Mikhail Gelfand was a chair of the program committee and a deputy chair of the local organizing committee, Maria Poptsova was also a member of the latter.

On August 21, the ninth workshop What Can FCA Do for Artificial Intelligence? took place, organised by Prof. Sergei Kuznetsov together with Université de Lorraine (Amedeo Napoli), and Technische Universität Dresden (Sebastian Rudolph). The workshop was co-located with IJCAI 2021 (A*). Prof. Kuznetsov, Dmitry Ignatov, Elizaveta Goncharova, Egor Dudyrev, and Ilya Semenkov gave their talks at the workshop.

The second international Summer School on Machine Learning in Bioinformatics took place on August 23–27. This year’s school featured 533 participants from 53 countries. Topics of the Summer School included machine learning applications in genetics and proteomics, graph neural networks for protein 3D structures, and bayesian checkpoint segmentation and genome annotation. Maria Poptsova gave one of the lectures, assisted by Nazar Beknazarov.

Full programme

Playlist

On September, 3-4, the first event of the Genetics of Cardiovascular Diseases consortium — an interdisciplinary forum Cardiogenetics and Bioinformatics: Realities 2021 — was held at HSE University. HSE University Vice-Rector Maria Yudkevich and Dean of the Faculty of Computer Science Prof. Ivan Arzhantsev spoke at the opening; while Maria Poptsova represented HSE University’s part of the consortium.

On October 25-29, the international conference Topology of Torus Actions and Related Topics was held at the Sirius Mathematics Center (Sochi, Russia), organised by the International Laboratory of Algebraic Topology and Its Applications.

Full programme

This conference was followed by the international school Toric Topology and Combinatoric on November 1-5.

Lectures and workshops

On October 26-30, Statistics, Artificial Intelligence, Machine Learning, Probability, Learning Theory Event (SAMPLE) took place in Gelendzhik, Russia. The speakers of the conference were Profs. Vladimir Spokoiny and Denis Belomestny, students and alumni of the Math of Machine Learning master's programme, representatives of Skoltech and Huawei.

Lectures and workshops

On November 23-26, the third autumn school on generative models was organised by LAMBDA, Centre of Deep Learning and Bayesian Methods, and the International Laboratory of Stochastic Algorithms and High-Dimensional Inference (HDI Lab). Yandex was an industry partner of the school.

Lectures and workshops

On December 4-5, the Faculty hosted a Machine Learning Bootcamp to prepare the participants for ML competitions, including the IDAO 2022 International Data Analytics Olympiad. Students from more than 70 countries participated in the Bootcamp.

IDAO 2022 registration is open now!

Once again, the Faculty of Computer Science, Yandex, and the Platinum Partner Otritie Bank unite their efforts to challenge data scientists and machine learning experts from all over the world. The first task, provided by the LAMBDA lab, is about predicting the properties of two-dimensional crystals.

The application deadline is February 13.

On December 10-12, the conference Problems in Theoretical Computer Science 2021 took place. It was held by the Laboratory of Theoretical Computer Science.

On December 28, the international workshop Logic Matters (LM-2021) took place, organised by the School of Data Analysis and Artificial Intelligence.

Regular seminars

Faculty Colloquium

Playlist

• February 18    Of Points and Lines: Graphs, Metrics, and Betweenness (Laurent Beaudou, HSE University)
• March 2    Neural entity linking (Alexander Panchenko, Skoltech)
• April 13    Genetic computers (Alan Herbert, HSE University, InsideOutBio)
• May 18    Bruhat interval polytopes which are cubes (Mikiya Masuda, HSE University, Osaka City University Advanced Mathematical Institute)
• June 1    Computer Vision and Machine Learning for Environmental Monitoring (Konrad Schindler, ETH Zurich)
• September 28    Positional Embedding in Transformer-based Models (Tatiana Likhomanenko, Apple)
• October 26    Is the next deep learning disruption in the physical sciences? (Max Welling, University of Amsterdam)
• November 23    Software Release anomaly detection in DevOps environment (Manuel Mazzara, Innopolis University)
• December 7    Highly accurate protein structure prediction with AlphaFold (Anna Potapenko, DeepMind)

Bayesian Methods Research Group

• March 5    Stochastic gradient estimation for discrete variables (Denis Rakitin)
• March 12    Meta-learning in neural networks (Alexander Grishin)
• March 26 April 9    Tensor Programs (Evgeny Golikov)
• April 16    Learning differential equations that are easy to solve (Evgeny Egorov)
• April 23     Audio Synthesis and Bandwidth Extension (Aibek Alanov, Oleg Ivanov)
• April 30     Minibatch acceptance test for Metropolis-Hastings (Andrey Chernov)
• May 14    Imitation Learning from Observations (Artyom Tsipin)
• September 17, 24    Neural Program Synthesis (Nadia Chirkova, Sergey Troshin)
• October 1    Large-Scale Wasserstein Gradient Flows (Petr Mokrov)
• October 8, 15, 22    Diffusion models (Grigory Bartosh, Victor Oganesyan)
• October 29    The Machine Learning of Time and Dynamics … with an Outlook towards the Sciences (Efstratios Gavves)
• November 12    Highly accurate protein structure prediction with AlphaFold (Mikhael Figurnov, DeepMind)
• November 26    Controlling GANs Latent Space (Vadim Titov)
• December 3    From classical statistics to modern deep learning (Mikhail (Misha) Belkin)
• December 17    Resolution-robust Large Mask Inpainting with Fourier Convolutions (and a brief survey of image inpainting) (Roman Suvorov)

International Laboratory of Theoretical Computer Science

• March 10    Studies on Insertion-deletion Systems and Contextual Grammars Related to Ambiguity, Measures and Applications (Anand Mahendran)
• April 14    Optimal monomial quadratization for ODE systems (Andrey Bychkov)
• May 19    An Optimal Separation of Randomized and Quantum Query Complexity (Andrey Storozhenko)

International Laboratory of Stochastic Algorithms and High-Dimensional Inference

• February 9    Rates of convergence for density estimation with GANs (Nikita Puchkin)
• February 16    On the Stability of Random Matrix Product with Markovian Noise: Application to Linear Stochastic Approximation and TD Learning (Sergey Samsonov)
• March 2    Approximation with deep neural networks (Dmitry Yarotsky)
• March 9    Exponential Savings in Agnostic Active Learning through Abstention (Nikita Puchkin)
• March 16    Distribution-free Robust Linear Regression (Nikita Zhivotovski)
• March 23    Stochastic normalizing flows (Maxim Panov)
• March 30    Constructive Universal High-Dimensional Distribution Generation through Deep ReLU (Dmytro Perekrestenko)
• April 6    Improved Complexity Bounds in the Wasserstein Barycenter Problem (Daniil Tyapkin)
• April 13    Robust k-means clustering in metric spaces (Alexey Kroshnin)
• April 20    Online learning with exponential weights in metric spaces (Quentin Paris)
• May 18    Fast rates for strongly convex optimization via stability (Egor Klotchkov)
• June 1     UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms (Sergey Samsonov)

International Laboratory of Algebraic Topology and Its Applications

• May 13    A Higher-Dimensional Generalization of the Heawood Theorem on Embeddings of Graphs into Surfaces (Eugene Kogan and Arkadiy Skopenkov)
• May 28    Hodge Laplacian via Random Walks on Simplicial Complexes (Gabor Lippner)
• June 4    Metric and Topological Approaches to Network Data Analysis (Samir Chowdhury)
• November 12 Topological Data Analysis can be enough for Classification (Rolando Kindelan Nuñez)

Laboratory of Complex Systems Modeling and Control

• January 22    Keeping Your Friends Close: Land Allocation with Friends (Edith Elkind)
• April 15    Historical datasets- implications for space weather risk over multiple solar cycles (Sandra Chapman)
• November 11    Existence of Extreme event: Understanding the mechanism via simple complex system (Sabarathinam Srinivasan)

Laboratory of Process-Aware Information Systems (PAIS Lab)

• February 13, 20    Petri Net Transformations: Abstraction, Refinement, Morphisms (Roman Nesterov)
• 5 марта High-level Petri net dynamic visualization. First step (Yaroslav Kotylev)
• 22 апреля, 6 мая Efficient equivalence checking technique for some classes of automata (Vladimir Zakharov)
• 22 мая Modeling and analysis of processes in protocol systems using Coloured Petri Nets: study case IEEE 802.16 standard (Ana Morales Bezeira)
• 17 июня Object-Centric Replay-Based Conformance Checking: Unveiling Desire Lines and Local Deviations (Julio Carrasquel)
• 29 июня Data and Reference Semantic-Based Simulator of DB-nets with the Use of Renew Too (Anton Rigin)
• September 8    Studying the properties of timed finite state machines (Evgenii Vinarskii)
• September 15    Overview of real-time models (Evgenii Vinarskii)

Laboratory on AI for Computational Biology

• June 25, 29    Introduction to computational mass spectrometry (Attila Kertesz-Farkas)
• October 1    Deep Convolutional Neural Networks Help Scoring Tandem Mass Spectrometry Data in Database-Searching Approaches (Polina Kudryavtseva)
• November 26    Computational methods for tandem mass spectrometry data annotation (Attila Kertesz-Farkas)
• December 17    Neuro-Symbolic approaches in Visual Question Answering problems (Insan-Aleksandr Latypov)

New departments and projects

This year, two laboratories have been set up at the Faculty of Computer Science which focus on the application of computer science in biology. The International Laboratory of Bioinformatics opened in January. It succeeds the previous Bioinformatics Lab. It is headed by Maria Poptsova, with Alan Herbert, founder of InsideOutBio and discoverer of Z-DNA, the left-handed variant of the DNA molecule, as a research advisor.

The Laboratory on AI for Computational Biology opened in March. It is headed by Attila Kertes-Farkas. The lab will develop deep learning technologies for analysing biomolecular medical data.

The Faculty also has two new departments. The Otkritie Bank Department was created in January and is headed by Sergey Rusanov, a member of the management board at Otkritie Bank. The JetBrains Department was created as a part of the School of Software Engineering in April, its head is Alexey Mitsyuk.

In early October HSE University won the grant from the Government of Russia to create an AI research centre as part of the federal Artificial Intelligence project. The Faculty of Computer Science will play a leading part in the creation of the HSE University Artificial Intelligence Research Centre, along with other units of the university.

The Laboratory of Methods for Big Data Analysis (LAMBDA) partakes in constructing an electron-positron collider to conduct comprehensive research into the charm quarks of D mesons and tau leptons. Financed by the Government of Russia, the project calls for research teams of HSE University, Russian Academy of Sciences, Yandex School of Data Analysis, and Rosatom.

 

Our congratulations to the graduates of the Faculty's Doctoral School who defended their theses this year: Vladislav Shakhuro, Soroosh Shalileh, Nikolay Smetanin, Anvar Kurmukov, and Dmitry Altukhov. Liudmila Prokhorenkova, who received her habilitation and the degree of Doctor of Sciences this year, deserves a special mention.

We also congratulate Prof. Dmitry Vetrov who received the Research Excellence Award Russia 2021, and Egor Kosov who received the Moscow Mathematical Society Award in November.