Alexander Andreevich Karandeev
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Alexander Andreevich Karandeev has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Ilya Makarov
AXXX, Moscow, Russian Federation and Trusted AI Center, RAS, Moscow, Russian Federation
3 shared publications - Victor Ivanovich Baluta
Keldysh Institute of Applied Mathematics of the Russian Academy of Sciences
3 shared publications - Ahmad Ahmad
MIPT, Moscow, Russia
2 shared publications - Aleksandr Kovalenko
AIRI, Moscow, Russian Federation
2 shared publications - Alexey Maslov
Artificial Intelligence Research Center, National Research Nuclear University MEPhI, Moscow, Russia
2 shared publications - Nikita Andreevich Yashin
Plekhanov Russian University of Economics
2 shared publications
Latest Publications
- Open Access
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 213753-213768
BinaryCAN: Learning Compact Binary Embeddings From Automotive CAN Bus Data
Modern vehicles generate high-frequency, multivariate data streams from the Controller Area Network (CAN) bus, the fundamental digital nervous system for in-vehicle communication. Although its primary role is real-time control of vehicle subsystems, the CAN bus also serves as a rich source of derived information on vehicle state and driver behavior. These data are particularly valuable for Vehicle-to-Everything (V2X) applications such as driver identification, predictive maintenance, and cooperative perception. However, transmitting raw data is often impractical due to bandwidth limitations in wireless networks. To address this challenge, we propose BinaryCAN, a novel framework that learns compact latent representations directly from automotive time series using autoencoders with binary bottlenecks. Our approach leverages the Straight-Through Estimator (STE) to enforce discrete embeddings while preserving task-relevant semantics.We evaluated three architectures, including a new binary encoder based on the graph neural network (GNN), on downstream tasks such as fuel supplier classification, oil brand identification, speed regression, and driver fingerprinting. Experimental results demonstrate that our learned embeddings not only maintain strong performance in these tasks while achieving substantial compression compared to the raw data, but in some cases even surpass raw data performance. Our work suggests that learned binary embeddings can potentially complement or enhance traditional V2X messaging by enabling more efficient transmission of semantically rich information.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 213753-213768
Open AccessBinaryCAN: Learning Compact Binary Embeddings From Automotive CAN Bus Data
Modern vehicles generate high-frequency, multivariate data streams from the Controller Area Network (CAN) bus, the fundamental digital nervous system for in-vehicle communication. Although its primary role is real-time control of vehicle subsystems, the CAN bus also serves as a rich source of derived information on vehicle state and driver behavior. These data are particularly valuable for Vehicle-to-Everything (V2X) applications such as driver identification, predictive maintenance, and cooperative perception. However, transmitting raw data is often impractical due to bandwidth limitations in wireless networks. To address this challenge, we propose BinaryCAN, a novel framework that learns compact latent representations directly from automotive time series using autoencoders with binary bottlenecks. Our approach leverages the Straight-Through Estimator (STE) to enforce discrete embeddings while preserving task-relevant semantics.We evaluated three architectures, including a new binary encoder based on the graph neural network (GNN), on downstream tasks such as fuel supplier classification, oil brand identification, speed regression, and driver fingerprinting. Experimental results demonstrate that our learned embeddings not only maintain strong performance in these tasks while achieving substantial compression compared to the raw data, but in some cases even surpass raw data performance. Our work suggests that learned binary embeddings can potentially complement or enhance traditional V2X messaging by enabling more efficient transmission of semantically rich information.
- Open Access
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1404-1406
Poster: Car2Vec: Task-Agnostic Latent Embeddings from CAN Bus for Efficient V2X Communication
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1404-1406
Open AccessPoster: Car2Vec: Task-Agnostic Latent Embeddings from CAN Bus for Efficient V2X Communication
- Open Access
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 170106-170115
LCRM: A Linear-Camera Dataset for Road Marking Segmentation and Damage Evaluation
We introduce the Linear-Camera Road Marking (LCRM) dataset, which is the first to couple segmentation with marking damage evaluation on the same images, and the first captured with a linear (line-scan) camera. LCRM spans Russian cities, covers 12 marking types, and grades instance damage with a 3-level scheme aligned with maintenance thresholds. We release 3,966 labeled images (train/test) and 18,847 unlabeled images. We also benchmark representative models across four tasks (binary and multi-class semantic segmentation, instance segmentation, and damage evaluation) to establish baselines for this new sensing regime.
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 170106-170115
Open AccessLCRM: A Linear-Camera Dataset for Road Marking Segmentation and Damage Evaluation
We introduce the Linear-Camera Road Marking (LCRM) dataset, which is the first to couple segmentation with marking damage evaluation on the same images, and the first captured with a linear (line-scan) camera. LCRM spans Russian cities, covers 12 marking types, and grades instance damage with a 3-level scheme aligned with maintenance thresholds. We release 3,966 labeled images (train/test) and 18,847 unlabeled images. We also benchmark representative models across four tasks (binary and multi-class semantic segmentation, instance segmentation, and damage evaluation) to establish baselines for this new sensing regime.
- Open Access
- 1 December 2023
- Published by National Research Nuclear University MEPhI (Moscow Engineering Physics Institute) in Scientific Visualization
- Vol. 15 (5) , 64-72
Visualization of a graph model of medical drugs and their classes with subject to incompatibility
Currently, the world is actively undergoing digital transformation, which is associated with the introduction of digital technologies in various fields. Methods for predicting and combating various types of diseases are being developed more and more actively. A separate place in this issue is given to pharmacology and reducing the risks of poly-pharmacotherapy. Works in this area have become more and more active in the last three years, and relate primarily to the problems of experimental medicine - the search for new drugs or classification of patients by images. In this connection, the development of algorithms for processing big data in the field of clinical pharmacology, including regulatory documents, instructions for the use of drugs, reliable medical resources for machine learning and building a knowledge base focused on practical application in real medical practice, is becoming increasingly important. One of the considered approaches for solving the problems described above is the search for an effective vector representation of the found concepts (word embedding) to build models of recommendations for the practical use of drugs based on the diagnosis and models for identifying their compatibility. The article presents a software tool for visualizing a graph model of medicines and their classes, considering their incompatibility. The problem of incompatibility of medicines is being investigated. The considered database of medicines, obtained from various sources, and the advantages of its use are described. The article presents the results of the developed prototype of a software tool that allows you to visualize a complete graph of relationships between drugs, a graph of incompatible drugs for a selected drug, and a graph of compatibility of two drugs. Graph model visualization provides a visual representation of complex relationships and allows researchers and pharmacologists to better understand interactions between drugs. The software offer the necessary tools to make informed decisions when prescribing medications and improve safety when using medicines.
- 1 December 2023
- Published by National Research Nuclear University MEPhI (Moscow Engineering Physics Institute) in Scientific Visualization
- Vol. 15 (5) , 64-72
Open AccessVisualization of a graph model of medical drugs and their classes with subject to incompatibility
Currently, the world is actively undergoing digital transformation, which is associated with the introduction of digital technologies in various fields. Methods for predicting and combating various types of diseases are being developed more and more actively. A separate place in this issue is given to pharmacology and reducing the risks of poly-pharmacotherapy. Works in this area have become more and more active in the last three years, and relate primarily to the problems of experimental medicine - the search for new drugs or classification of patients by images. In this connection, the development of algorithms for processing big data in the field of clinical pharmacology, including regulatory documents, instructions for the use of drugs, reliable medical resources for machine learning and building a knowledge base focused on practical application in real medical practice, is becoming increasingly important. One of the considered approaches for solving the problems described above is the search for an effective vector representation of the found concepts (word embedding) to build models of recommendations for the practical use of drugs based on the diagnosis and models for identifying their compatibility. The article presents a software tool for visualizing a graph model of medicines and their classes, considering their incompatibility. The problem of incompatibility of medicines is being investigated. The considered database of medicines, obtained from various sources, and the advantages of its use are described. The article presents the results of the developed prototype of a software tool that allows you to visualize a complete graph of relationships between drugs, a graph of incompatible drugs for a selected drug, and a graph of compatibility of two drugs. Graph model visualization provides a visual representation of complex relationships and allows researchers and pharmacologists to better understand interactions between drugs. The software offer the necessary tools to make informed decisions when prescribing medications and improve safety when using medicines.
- Open Access
- 1 January 2023
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 57,p. 1-12
Development of a drug interaction assessment model
The work proposes an approach for assessing compatibility between drugs. The method under consideration is constructed using cluster analysis In the course of research, medical drugs are grouped in threes with consideration of interaction between them, which allows to systematize hundreds, and sometimes thousands of drugs.
- 1 January 2023
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 57,p. 1-12
Open AccessDevelopment of a drug interaction assessment model
The work proposes an approach for assessing compatibility between drugs. The method under consideration is constructed using cluster analysis In the course of research, medical drugs are grouped in threes with consideration of interaction between them, which allows to systematize hundreds, and sometimes thousands of drugs.
- 1 January 2021
- Published by Keldysh Institute of Applied Mathematics
Application of the boundary detection method to the problem of recognizing the situation
This paper presents the results of one of the solutions to the problem of increasing the speed of decision-making by an intelligent agent when modeling the behavior of complex systems on a virtual electronic polygon. Such a training ground is currently considered as an instrumental platform for testing technologies for training intelligent agents in conditions of varying complexity in order to subsequently transfer the developed methods to real objects for solving practical problems. As an example, the control of a robotic device operating in an enclosed space is considered. The article describes the technology of reducing the volume and dimension of the processed data in order to increase the responsiveness to changes in the situation and the development of solutions for moving a robotic device. The technology is based on the preprocessing of video images for the formation of a training sample, as well as the procedure and results of deep learning of a convolutional neural network. The paper uses an open source library of OpenCV computer vision algorithms implemented in C / C++. It is shown that focusing on the selection of object boundaries can significantly reduce the amount of data for analyzing the situation and increase the speed of decision-making by the robot to move.
- 1 January 2021
- Published by Keldysh Institute of Applied Mathematics
Application of the boundary detection method to the problem of recognizing the situation
This paper presents the results of one of the solutions to the problem of increasing the speed of decision-making by an intelligent agent when modeling the behavior of complex systems on a virtual electronic polygon. Such a training ground is currently considered as an instrumental platform for testing technologies for training intelligent agents in conditions of varying complexity in order to subsequently transfer the developed methods to real objects for solving practical problems. As an example, the control of a robotic device operating in an enclosed space is considered. The article describes the technology of reducing the volume and dimension of the processed data in order to increase the responsiveness to changes in the situation and the development of solutions for moving a robotic device. The technology is based on the preprocessing of video images for the formation of a training sample, as well as the procedure and results of deep learning of a convolutional neural network. The paper uses an open source library of OpenCV computer vision algorithms implemented in C / C++. It is shown that focusing on the selection of object boundaries can significantly reduce the amount of data for analyzing the situation and increase the speed of decision-making by the robot to move.
- 1 January 2019
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 102,p. 1-18
Assessment of anti-terrorism security of objects based on the calculation of integrated indicators
In this paper, we propose an approach that allows us to obtain a preliminary assessment of the effectiveness of the protection system based on the use of a small number of input parameters by calculating the zoning of the object. The proposed calculation method can also be used to optimize the placement of system elements, taking into account the existing limitations.
- 1 January 2019
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 102,p. 1-18
Assessment of anti-terrorism security of objects based on the calculation of integrated indicators
In this paper, we propose an approach that allows us to obtain a preliminary assessment of the effectiveness of the protection system based on the use of a small number of input parameters by calculating the zoning of the object. The proposed calculation method can also be used to optimize the placement of system elements, taking into account the existing limitations.
- 1 January 2018
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 86,p. 1-22
Simulation of antagonistic conflicts in the third-order cybernetics paradigm
- 1 January 2018
- Published by Keldysh Institute of Applied Mathematics in Keldysh Institute Preprints
- No. 86,p. 1-22
Simulation of antagonistic conflicts in the third-order cybernetics paradigm