Mikhail Drobyshevskiy
Mikhail Drobyshevskiy has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Denis Yurievich Turdakov
Ivannikov Institute for System Programming of the Russian Academy of Sciences
4 shared publications Denis Turdakov
Ivannikov Institute for System Programming of the Russian Academy of Sciences and Moscow Institute of Physics and Technology, Moscow, Russia
2 shared publicationsA.A. Avetisyan
Ivannikov Institute for System Programming of the RAS, 25, Alexander Solzhenitsyn Str., Moscow, 109004, Russia
1 shared publication- Aleksandr Kovalenko
AIRI, Moscow, Russian Federation
1 shared publication - Andrey Gomzin
Ivannikov, Institute for System Programming, Lomonosov Moscow State University, Moscow, Russia
1 shared publication Anton Korshunov
Institute for System Programming of Russian Academy of Sciences
1 shared publication
Latest Publications
- Open Access
- 15 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Industrial Electronics Society
- Vol. 5, 428-440
Adversarial Attacks and Defenses in Fault Detection and Diagnosis: A Comprehensive Benchmark on the Tennessee Eastman Process
Integrating machine learning into Automated Control Systems (ACS) enhances decision-making in industrial process management. One of the limitations to the widespread adoption of these technologies in industry is the vulnerability of neural networks to adversarial attacks. This study explores the threats in deploying deep learning models for Fault Detection and Diagnosis (FDD) in ACS using the Tennessee Eastman Process dataset. By evaluating three neural networks with different architectures, we subject them to six types of adversarial attacks and explore five different defense methods. Our results highlight the strong vulnerability of models to adversarial samples and the varying effectiveness of defense strategies. We also propose a new defense strategy based on combining adversarial training and data quantization. This research contributes several insights into securing machine learning within ACS, ensuring robust FDD in industrial processes.
- 15 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Industrial Electronics Society
- Vol. 5, 428-440
Open AccessAdversarial Attacks and Defenses in Fault Detection and Diagnosis: A Comprehensive Benchmark on the Tennessee Eastman Process
Integrating machine learning into Automated Control Systems (ACS) enhances decision-making in industrial process management. One of the limitations to the widespread adoption of these technologies in industry is the vulnerability of neural networks to adversarial attacks. This study explores the threats in deploying deep learning models for Fault Detection and Diagnosis (FDD) in ACS using the Tennessee Eastman Process dataset. By evaluating three neural networks with different architectures, we subject them to six types of adversarial attacks and explore five different defense methods. Our results highlight the strong vulnerability of models to adversarial samples and the varying effectiveness of defense strategies. We also propose a new defense strategy based on combining adversarial training and data quantization. This research contributes several insights into securing machine learning within ACS, ensuring robust FDD in industrial processes.
- Open Access
- 1 January 2022
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 34 (5) , 127-142
Detecting Influential Users in Social Networks Based on Bipartite Comments Graph
С развитием онлайновых социальных сетей все большую актуальность приобретают задачи выявления пользователей, которые оказывает большое влияние на других участников социальных сетей. Важным источником информации служат данные о комментировании пользователями контента, создаваемого другими пользователями. В работе предлагается метод определения влиятельности, основанный на двудольном графе пользователь-комментарий-контент, а также использующий информацию о текстовых сообщениях и реакции на них других пользователей. Кроме того, предложен метод выявления сообществ пользователей в таком графе на основе общих интересов. Результаты тестирования на коллекциях данных сетей ВКонтакте и YouTube показывают корреляцию активности и влиятельности пользователя, однако самые активные комментаторы не обязательно являются самыми влиятельными. Анализ сообществ показывает положительную корреляцию между размером сообщества, числом наиболее влиятельных пользователей в нем и средней влиятельностью пользователей сообщества.
- 1 January 2022
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 34 (5) , 127-142
Open AccessDetecting Influential Users in Social Networks Based on Bipartite Comments Graph
С развитием онлайновых социальных сетей все большую актуальность приобретают задачи выявления пользователей, которые оказывает большое влияние на других участников социальных сетей. Важным источником информации служат данные о комментировании пользователями контента, создаваемого другими пользователями. В работе предлагается метод определения влиятельности, основанный на двудольном графе пользователь-комментарий-контент, а также использующий информацию о текстовых сообщениях и реакции на них других пользователей. Кроме того, предложен метод выявления сообществ пользователей в таком графе на основе общих интересов. Результаты тестирования на коллекциях данных сетей ВКонтакте и YouTube показывают корреляцию активности и влиятельности пользователя, однако самые активные комментаторы не обязательно являются самыми влиятельными. Анализ сообществ показывает положительную корреляцию между размером сообщества, числом наиболее влиятельных пользователей в нем и средней влиятельностью пользователей сообщества.
- 24 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Context Specificity Matters: Profile Attributes Prediction for Social Network Users
Analysis of social network user profiles plays an important role in various applications such as recommender systems and marketing services. In this paper we consider the task of profile prediction given a sample of a social graph. We introduce a new node property named context specificity, which allows to enhance node attribute prediction methods by emphasizing more useful nodes and edges. Particularly, we show how to apply it to several existing methods: label propagation, classification based on node embeddings, graph convolutional neural network. Our experiments prove that context specificity feature improves methods quality in most considered tasks. Additionally, we suggest a new supervised method based on attribute values distributions that is competitive with state of the art approaches. We also describe and share two new datasets, representing a sample of VK social network.
- 24 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Context Specificity Matters: Profile Attributes Prediction for Social Network Users
Analysis of social network user profiles plays an important role in various applications such as recommender systems and marketing services. In this paper we consider the task of profile prediction given a sample of a social graph. We introduce a new node property named context specificity, which allows to enhance node attribute prediction methods by emphasizing more useful nodes and edges. Particularly, we show how to apply it to several existing methods: label propagation, classification based on node embeddings, graph convolutional neural network. Our experiments prove that context specificity feature improves methods quality in most considered tasks. Additionally, we suggest a new supervised method based on attribute values distributions that is competitive with state of the art approaches. We also describe and share two new datasets, representing a sample of VK social network.
- 16 December 2019
- Published by Pleiades Publishing Ltd in Programming and Computer Software
- Vol. 45 (7) , 372-380
Methods for Information Diffusion Analysis
Information diffusion is a fundamental process taking place on the Internet. Every day, various pieces of information are published and further disseminated through news articles and user...
- 16 December 2019
- Published by Pleiades Publishing Ltd in Programming and Computer Software
- Vol. 45 (7) , 372-380
Methods for Information Diffusion Analysis
Information diffusion is a fundamental process taking place on the Internet. Every day, various pieces of information are published and further disseminated through news articles and user...
- Open Access
- 10 December 2019
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 52 (6) , 1-36
Random Graph Modeling
Random graph (RG) models play a central role in complex networks analysis. They help us to understand, control, and predict phenomena occurring, for instance, in social networks, biological networks, the Internet, and so on.Despite a large number of RG models presented in the literature, there are few concepts underlying them. Instead of trying to classify a wide variety of very dispersed models, we capture and describe concepts they exploit considering preferential attachment, copying principle, hyperbolic geometry, recursively defined structure, edge switching, Monte Carlo sampling, and so on. We analyze RG models, extract their basic principles, and build a taxonomy of concepts they are based on. We also discuss how these concepts are combined in RG models and how they work in typical applications like benchmarks, null models, and data anonymization.
- 10 December 2019
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 52 (6) , 1-36
Open AccessRandom Graph Modeling
Random graph (RG) models play a central role in complex networks analysis. They help us to understand, control, and predict phenomena occurring, for instance, in social networks, biological networks, the Internet, and so on.Despite a large number of RG models presented in the literature, there are few concepts underlying them. Instead of trying to classify a wide variety of very dispersed models, we capture and describe concepts they exploit considering preferential attachment, copying principle, hyperbolic geometry, recursively defined structure, edge switching, Monte Carlo sampling, and so on. We analyze RG models, extract their basic principles, and build a taxonomy of concepts they are based on. We also discuss how these concepts are combined in RG models and how they work in typical applications like benchmarks, null models, and data anonymization.
- Open Access
- 1 January 2019
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 31 (5) , 137-144
Methods for News Items Popularity Estimation on Early Stages
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- 1 January 2019
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 31 (5) , 137-144
Open AccessMethods for News Items Popularity Estimation on Early Stages
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- 30 December 2017
- Published by Springer Nature
- p. 634-650
Learning and Scaling Directed Networks via Graph Embedding
- 30 December 2017
- Published by Springer Nature
- p. 634-650
Learning and Scaling Directed Networks via Graph Embedding