• A
  • A
  • A
  • АБВ
  • АБВ
  • АБВ
  • A
  • A
  • A
  • A
  • A
Обычная версия сайта
  • RU
  • EN
  • HSE University
  • Publications
  • Articles
  • Predicting Molecule Toxicity via Descriptor-based Graph Self-supervised Learning
  • RU
  • EN
Расширенный поиск
Высшая школа экономики
Национальный исследовательский университет
Priority areas
  • business informatics
  • economics
  • engineering science
  • humanitarian
  • IT and mathematics
  • law
  • management
  • mathematics
  • sociology
  • state and public administration
by year
  • 2027
  • 2026
  • 2025
  • 2024
  • 2023
  • 2022
  • 2021
  • 2020
  • 2019
  • 2018
  • 2017
  • 2016
  • 2015
  • 2014
  • 2013
  • 2012
  • 2011
  • 2010
  • 2009
  • 2008
  • 2007
  • 2006
  • 2005
  • 2004
  • 2003
  • 2002
  • 2001
  • 2000
  • 1999
  • 1998
  • 1997
  • 1996
  • 1995
  • 1994
  • 1993
  • 1992
  • 1991
  • 1990
  • 1989
  • 1988
  • 1987
  • 1986
  • 1985
  • 1984
  • 1983
  • 1982
  • 1981
  • 1980
  • 1979
  • 1978
  • 1977
  • 1976
  • 1975
  • 1974
  • 1973
  • 1972
  • 1971
  • 1970
  • 1969
  • 1968
  • 1967
  • 1966
  • 1965
  • 1964
  • 1963
  • 1958
  • More
Subject
News
November 17, 2026
Physicists Show How a Solvent Shapes the Response of Drug Molecules
Many drug molecules can switch between different three-dimensional shapes. Using a mathematical model of ibuprofen dissolved in supercritical CO₂, researchers from HSE University and the Institute of Solution Chemistry of the Russian Academy of Sciences have shown that the solvent can make these molecules appear to respond unusually slowly. The molecules themselves continue to change shape rapidly, but their surroundings rearrange much more slowly, and this lag shows up in the measurements. The study has been published in The Journal of Chemical Physics. 
November 10, 2026
Physicists Train AI to Optimise Particle Beams in Accelerators
Researchers from the AI and Digital Science Institute at the HSE Faculty of Computer Science and the Budker Institute of Nuclear Physics at the Siberian Branch of the Russian Academy of Sciences have developed RLABC (Reinforcement Learning for Accelerator Beamline Control), a framework that uses reinforcement learning to identify optimal settings for a particle accelerator. In a computer simulation, RLABC optimised the accelerator's parameters to achieve a particle transmission rate of 70%. The study has been published in Computer Physics Communications.
November 9, 2026
'Organ-on-a-Chip Technology Could Eliminate the Need for Animal Testing'
The Laboratory of Molecular Physiology at the HSE Faculty of Biology and Biotechnology investigates how molecular regulators known as microRNAs control gene expression and contribute to the development of diseases, including cancer, viral and inflammatory conditions, and pregnancy complications. At HSE University, researchers develop organ-on-a-chip models: miniature devices containing living cells that mimic the functions of human organs, such as the placenta and intestines. These models help researchers understand how diseases develop and how drugs affect them. HSE News Service spoke with Evgeny Knyazev, Head of the Laboratory.

 

Have you spotted a typo?
Highlight it, click Ctrl+Enter and send us a message. Thank you for your help!

Publications
  • Books
  • Articles
  • Chapters of books
  • Working papers
  • Report a publication
  • Research at HSE

?

Predicting Molecule Toxicity via Descriptor-based Graph Self-supervised Learning

IEEE Access. 2023. Vol. 11. P. 91842–91849.
Li X., Makarov I., Kiselev D.

Predicting molecular properties with Graph Neural Networks (GNNs) has recently drawn a lot of attention, with compound toxicity prediction being one of the biggest challenges. In cases where there is insufficient labeled molecule data, an effective approach is to pre-train GNNs on large-scale unlabeled molecular data and then fine-tune them for downstream tasks. Among pre-training strategies, node-level pre-training involves masking and predicting atom properties, while motif-based methods capture rich information in subgraphs. These approaches have shown effectiveness across various downstream tasks. However, current pre-training frameworks face two main challenges: (1) node-level auxiliary tasks do not preserve useful domain knowledge, and (2) the fusion of motif-based methods and node-level tasks is computationally extensive. To address these challenges, we propose Descriptor-based Graph Self-supervised Learning (DGSSL), a method that utilizes domain knowledge to enhance graph representation learning. Specifically, it identifies descriptor centers in molecules and encodes motif-like information as special atomic numbers in the pre-training tasks. This enables node-level self-supervised pre-training frameworks for GNNs to also capture rich information in local subgraphs. Experimental results demonstrate that our method achieves state-of-the-art performance on three toxicity-related benchmarks.

Research target: Computer Science
Language: English
DOI
Text on another site
Keywords: Self-supervised learning Graph neural networksmolecule toxicity prediction
Publication based on the results of:
Models and method for analysis of unstructured data, data mining and recommender systems (2023)
Similar publications
Agreement among large language models does not track medical truth in health claim verification
Artemenko E., Koltsova O., Nikolenko S. et al., Scientific Reports 2026
Large language models (LLMs) are increasingly proposed as tools to verify online health information. A common assumption from the self-consistency and ensemble literature is that agreement across models signals reliability. But it remains untested whether inter-model consensus tracks medical truth. We evaluated 10 LLMs spanning frontier, Russian-specific and medical-specialized families on RuHealth, a corpus of ...
Added: November 20, 2026
FLowRA: Flow-Guided Rank Allocation for Parameter-Efficient Fine-Tuning of Deep Neural Networks
Aleksei Samarin, Nazarenko A., Kotenko E. et al., Machine Learning and Knowledge Extraction 2026 Vol. 8 No. 11 Article 389
Low-rank adaptation (LoRA) is widely used for parameter-efficient fine-tuning, yet uniform rank assignment ignores the varying contribution of layers to a downstream task. Adaptive methods such as AdaLoRA allocate ranks via gradient and singular-value importance, but they do not exploit the representation change induced by the task. We introduce FLowRA, a calibration-based rank-allocation framework that ...
Added: November 19, 2026
Machine Learning for Astrophysics 2024
Springer, 2026.
This proceedings book reviews the state of the art in the exploitation of machine learning techniques for the astrophysics community and gives the reader a complete overview of the field. The contributed chapters allow the reader to easily digest the material through balanced theoretical and numerical methods and tools with applications in different fields of ...
Added: November 19, 2026
Approximation of Multivariate Time Series in Astronomy Using Vector Gaussian Processes
Semenikhin T. A., Kornilov M., Pruzhinskaya M. V., Moscow University Physics Bulletin 2025 Vol. 80 No. S1 P. S204–S207
In astronomy, it is often necessary to approximate sparse vector functions, such as multiband light curves of transient objects. Traditional methods typically treat each photometric band independently, ignoring interband correlations and thus failing to fully exploit the available multiband information in sparsely sampled datasets. We propose a vector Gaussian process (GP) framework that models multi-band ...
Added: November 19, 2026
Anomaly detection in Fink: I. Discovery, follow-up, and classification of unusual sources
Pruzhinskaya M. V., Kornilov M., Dodin A. V. et al., Astronomy and Astrophysics 2026 Vol. 715 P. 1–18
Context. Modern wide-field time-domain surveys produce alert streams whose scientific potential is often concentrated in rare and unusual events. Efficient discovery therefore requires automated pipelines to be combined with rapid expert validation and follow-up. Aims. We present the first-year performance of the anomaly detection (AD) pipeline operating within the Fink broker on the Zwicky Transient ...
Added: November 19, 2026
Supernova scores for active anomaly detection
Semenikhin T. A., Kornilov M.V., Pruzhinskaya M. V. et al., Astronomy and Computing 2026 Vol. 56 Article 101126
Large time-domain sky surveys generate extensive multi-year catalogs of light curves in which scientifically valuable transients, such as supernovae (SNe), are vastly outnumbered by artifacts and routine star variability. While supervised machine learning models can efficiently filter known classes, they struggle with extreme class imbalance and may overlook rare or novel events. Conversely, unsupervised anomaly ...
Added: November 19, 2026
Active learning of collinear magnetic Moment Tensor Potentials using the spin-MLIP package from soft-constrained spin-polarized DFT calculations: a case study of Fe-Pd
Burov A., Котыхов А. С., Aksyonov D. et al., Computer Physics Communications 2027 Vol. 330 Article 110458
Explicit incorporation of magnetic degrees of freedom in machine-learning interatomic potentials (magnetic MLIPs) plays a crucial role in the correct description of magnetic materials and their properties. An important ingredient for fitting of magnetic MLIPs is spin-polarized density functional theory (DFT) calculations with non-equilibrium magnetic moments, i.e. DFT calculations with constraints on magnetic moments. In this study, we ...
Added: November 17, 2026
Toward reliable false discovery rate control in classification problems under distribution shift
Kertesz-Farkas A., Andrey B., Plos One 2026 Vol. 21 No. 11 P. 1–20
Background Domain shifts or batch effects can significantly degrade the performance of machine learning-based classifiers in software production without any warning. This can particularly be problematic in medical applications which involve human lives. The shift changes the shape of the prediction score distribution of the test samples compared to that of the training samples; hence, decision ...
Added: November 17, 2026
Траектории ставок как сигнал сговора: LSTM-классификация аукционов государственных закупок
Ефимов К. Д., Журнал Новой экономической ассоциации 2026 № 72 С. 105–114
This study proposes a method for automatic detection of collusive behavior in electronic public procurement auctions based on analysis of the full bid trajectory. The empirical dataset comprises 508 auction histories from 2016–2020 collected from the RTS tender e-procurement platform; the positive class is labeled using case materials of the Federal Antimonopoly Service (FAS Russia), ...
Added: November 13, 2026
Model-Dependent Prompt-Component Ablation for Medical LLM Decision Support: Application to ENT Biologics Screening
Sergey N. Koltcov, Budkovaia M., Surkov A. et al., IEEE Journal of Biomedical and Health Informatics 2026 P. 1–10
Large language models (LLMs) show promise for clinical decision support, but privacy constraints and unreliable free-text decisions limit deployment. We present Medotor, a locally deployed system for screening patients with chronic rhinosinusitis with nasal polyps (CRSwNP) for biologic therapy, and evaluate prompt design on 163 retrospective cases (132 eligible and 31 ineligible). Five models were ...
Added: November 13, 2026
Impact of 5G NR-V2X Message Losses on Connected Vehicle Warning Functions
Физулин А. В., Stepanyants V., Amerikanov A. et al., FUTURE TRANSPORTATION 2026 Vol. 6 No. 6 Article 264
Cooperative warning functions depend on whether V2X information becomes available in time to influence vehicle control. This article presents an event-based co-simulation framework coupling SUMO, ns-3 (ms-van3t), 5G-LENA NR sidelink and application records, with Sionna RT enabled for the intersection scenario. In 140 emergency-vehicle-warning runs, reducing mean packet reception ratio (PRR) from 0.9412 to 0.0675 ...
Added: November 12, 2026
Structural model of decision support system for sustainable development of oil and gas companies
Stoianova A. D., Trofimets V. Y., Matrokhina K. V., International Journal of Engineering, Transactions A: Basics 2025 Vol. 38 No. 4 P. 701–709
The article considers the issue of managerial decision-making in an oil and gas company, which is related to the class of multi-criteria choice problems. To improve the validity of decisions in the context of companies’ sustainable development it is proposed to implement a Decision Support System (DSS). A brief review of the existing methods of ...
Added: November 10, 2026
EdFitter: a hybrid course recommendation system using large language models
Sergei Koltcov, Ryumin D., Alexandr Axyonov et al., PeerJ Computer Science 2026 Article 12:e4141
Educational institutions face increasing pressure to adapt their curricula to the rapidly evolving demands of the job market, and recommender systems can be important tools to guide students in selecting courses that align with their career goals. We present EdFitter, a hybrid course recommendation system that combines pretrained text embeddings for retrieval with large language ...
Added: November 10, 2026
Assessing density equilibration in confined Liquid–Liquid interfaces
Guskov T., Gerke K., Kondratyuk N. et al., Molecular Simulation 2026 Vol. 52 No. 7 P. 546 – 555
Equilibration is often assumed rather than tested in molecular dynamics studies of confined liquid–liquid interfaces, even though different physical properties may relax on different time scales. In this work, equilibration is examined using an automated truncation method based on the marginal standard error rule. Interfacial geometry, wetting behaviour, and density-related properties are analysed for liquid–liquid ...
Added: November 10, 2026
Influence of nuclear quantum effects on metastable states of warm dense hydrogen
Lukyanchuk V., Norman G., Kondratyuk N. et al., Scientific Reports 2026 Vol. 16 No. 1 Article 16435
Previously, it has been effectively shown that nuclear quantum effects (NQE) considerably influence the prediction of the warm dense hydrogen phase diagram. Here, we study the influence of NQE on the parameters of the conducting state formation in warm dense hydrogen using path integral molecular dynamics (PIMD) within the framework of the finite temperature density ...
Added: November 10, 2026
Overcoming the Curse of Dimensionality with Synolitic AI
Zaikin A., Sviridov I., Sosedka A. et al., Technologies 2026 Vol. 14 No. 2 Article 84
High-dimensional tabular data are common in biomedical and clinical research, yet conventional machine learning methods often struggle in such settings due to data scarcity, feature redundancy, and limited generalization. In this study, we systematically evaluate Synolitic Graph Neural Networks (SGNNs), a framework that transforms high-dimensional samples into sample-specific graphs by training ensembles of low-dimensional pairwise ...
Added: May 23, 2026
Analysis of Multidimensional Clinical and Physiological Data with Synolitical Graph Neural Networks
Krivonosov M., Nazarenko T., Ushakov V. et al., Technologies 2025 Vol. 13 No. 1 Article 13
This paper introduces a novel approach for classifying multidimensional physiological and clinical data using Synolitic Graph Neural Networks (SGNNs). SGNNs are particularly good forto addressing the challenges posed by high-dimensional datasets, particularly in healthcare, where traditional mMachine lLearning and Artificial Intelligence methods often struggles to find global optima due to the “curse of dimensionality”. To ...
Added: December 23, 2024
Pose Networks Unveiled: Bridging the Gap for Monocular Depth Perception
Dayoub Y., Andrey V. Savchenko, Makarov I., , in: 2024 IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct).: IEEE, 2024. P. 584–587.
Depth estimation is essential in Augmented Reality applications, enabling realistic object placement, scene understanding, spatial mapping, interaction, and environment awareness. This paper proposes a method to enhance depth model performance without increasing inference costs by improving the pose network in a selfsupervised learning setup. In particular, we enrich spatial information in the pose network by ...
Added: December 3, 2024
SensorSCAN: Self-Supervised Learning and Deep Clustering for Fault Diagnosis in Chemical Processes
Maksim Golyadkin, Vitaliy Pozdnyakov, Leonid Zhukov et al., Artificial Intelligence 2023 Vol. 324 Article 104012
Modern industrial facilities generate large volumes of raw sensor data during the production process. This data is used to monitor and control the processes and can be analyzed to detect and predict process abnormalities. Typically, the data has to be annotated by experts in order to be used in predictive modeling. However, manual annotation of ...
Added: September 20, 2023
Adversarial attacks on graph-level embedding methods: a case study
Giordano M., Maddalena L., Manzo M. et al., Annals of Mathematics and Artificial Intelligence 2023 Vol. 91 P. 259–285
As the number of graph-level embedding techniques increases at an unprecedented speed, questions arise about their behavior and performance when training data undergo perturbations. This is the case when an external entity maliciously alters training data to invalidate the embedding. This paper explores the efects of such attacks on some graph datasets by applying diferent ...
Added: November 18, 2022
Exploration in Sequential Recommender Systems via Graph Representations
Kiselev D., Makarov I., IEEE Access 2022 Vol. 10 P. 123614–123621
Temporal graph networks are powerful tools for solving the cold-start problem in sequential recommender systems. However, graph models are susceptible to feedback loops and data distribution shifts. The paper proposes a simple yet efficient graph-based exploration method for the mitigation of the issues above. It adopts the counter-based state exploration from reinforcement learning to the ...
Added: September 5, 2022
Self-supervised recurrent depth estimation with attention mechanisms
Makarov I., Bakhanova M., Nikolenko S. et al., PeerJ Computer Science 2022 Vol. 8 Article e865
Depth estimation has been an essential task for many computer vision applications, especially in autonomous driving, where safety is paramount. Depth can be estimated not only with traditional supervised learning but also via a self-supervised approach that relies on camera motion and does not require ground truth depth maps. Recently, major improvements have been introduced ...
Added: February 1, 2022
On the Memorization Properties of Contrastive Learning
Sadrtdinov I., Chirkova N., Lobacheva E., , in: ICML 2021 Workshop, Overparameterization: Pitfalls & Opportunities.: [б.и.], 2021.
Memorization studies of deep neural networks (DNNs) help to understand what patterns and how do DNNs learn, and motivate improvements to DNN training approaches. In this work, we investigate the memorization properties of SimCLR, a widely used contrastive self-supervised learning approach, and compare them to the memorization of supervised learning and random labels training. We ...
Added: January 25, 2022
JONNEE: Joint Network Nodes and Edges Embedding
Makarov I., Korovina K., Kiselev D., IEEE Access 2021 Vol. 9 P. 144646–144659
Recently, graph embedding models significantly improved the quality of graph machine learning tasks, such as node classification and link prediction. In this work, we propose a model called JONNEE (JOint Network Nodes and Edges Embedding), which learns node and edge embeddings under self-supervision via joint constraints in a given graph and its edge-to-vertex dual representation ...
Added: October 30, 2021
  • About
  • About
  • Key Figures & Facts
  • Sustainability at HSE University
  • Faculties & Departments
  • International Partnerships
  • Faculty & Staff
  • HSE Buildings
  • HSE University for Persons with Disabilities
  • Public Enquiries
  • Studies
  • Admissions
  • Programme Catalogue
  • Undergraduate
  • Graduate
  • Exchange Programmes
  • Summer University
  • Summer Schools
  • Semester in Moscow
  • Business Internship
  • Research
  • International Laboratories
  • Research Centres
  • Research Projects
  • Monitoring Studies
  • Conferences & Seminars
  • Academic Jobs
  • Yasin (April) International Academic Conference on Economic and Social Development
  • Media & Resources
  • Publications by staff
  • HSE Journals
  • Publishing House
  • iq.hse.ru: commentary by HSE experts
  • Library
  • Economic & Social Data Archive
  • Video
  • HSE Repository of Socio-Economic Information
  • HSE1993–2026
  • Contacts
  • Copyright
  • Privacy Policy
  • Site Map
Edit