Showing 1–17 of 17 results for author: Selivanov, A

Global and Local Contrastive Learning for Joint Representations from Cardiac MRI and ECG

Authors: Alexander Selivanov, Philip Müller, Özgün Turgut, Nil Stolt-Ansó, Daniel Rückert

Abstract: An electrocardiogram (ECG) is a widely used, cost-effective tool for detecting electrical abnormalities in the heart. However, it cannot directly measure functional parameters, such as ventricular volumes and ejection fraction, which are crucial for assessing cardiac function. Cardiac magnetic resonance (CMR) is the gold standard for these measurements, providing detailed structural and functional… ▽ More An electrocardiogram (ECG) is a widely used, cost-effective tool for detecting electrical abnormalities in the heart. However, it cannot directly measure functional parameters, such as ventricular volumes and ejection fraction, which are crucial for assessing cardiac function. Cardiac magnetic resonance (CMR) is the gold standard for these measurements, providing detailed structural and functional insights, but is expensive and less accessible. To bridge this gap, we propose PTACL (Patient and Temporal Alignment Contrastive Learning), a multimodal contrastive learning framework that enhances ECG representations by integrating spatio-temporal information from CMR. PTACL uses global patient-level contrastive loss and local temporal-level contrastive loss. The global loss aligns patient-level representations by pulling ECG and CMR embeddings from the same patient closer together, while pushing apart embeddings from different patients. Local loss enforces fine-grained temporal alignment within each patient by contrasting encoded ECG segments with corresponding encoded CMR frames. This approach enriches ECG representations with diagnostic information beyond electrical activity and transfers more insights between modalities than global alignment alone, all without introducing new learnable weights. We evaluate PTACL on paired ECG-CMR data from 27,951 subjects in the UK Biobank. Compared to baseline approaches, PTACL achieves better performance in two clinically relevant tasks: (1) retrieving patients with similar cardiac phenotypes and (2) predicting CMR-derived cardiac function parameters, such as ventricular volumes and ejection fraction. Our results highlight the potential of PTACL to enhance non-invasive cardiac diagnostics using ECG. The code is available at: https://github.com/alsalivan/ecgcmr △ Less

Submitted 24 June, 2025; originally announced June 2025.

Comments: accepted to MICCAI 2025 (Springer LNCS)

Spillover-Free Control of the Semilinear Heat Equation via the $L^2$ Residue Separation and Harmonic Inequality

Authors: Anton Selivanov, Emilia Fridman

Abstract: We present a new method for designing spillover-free finite-dimensional output-feedback controllers for systems governed by parabolic partial differential equations (PDEs). Building on the recently developed $L^2$ residue separation for state-feedback control, we extend this concept to output-feedback using a novel harmonic inequality, which provides an optimal bound on the residue in terms of the… ▽ More We present a new method for designing spillover-free finite-dimensional output-feedback controllers for systems governed by parabolic partial differential equations (PDEs). Building on the recently developed $L^2$ residue separation for state-feedback control, we extend this concept to output-feedback using a novel harmonic inequality, which provides an optimal bound on the residue in terms of the weighted $l^2$ norm of its Fourier coefficients. We apply this approach to a 1D semilinear heat equation, where the measured output and unknown nonlinearity link the dominant and residual modes, leading to spillover if the residual modes are neglected in the controller design. We demonstrate how to compute the input-to-residue $L^2$ gain and utilize it to design a spillover-free controller. Additionally, we show how the refined approach eliminates the need for the lifting transformation, enabling a sample-and-hold implementation of the control signal. △ Less

Submitted 27 January, 2025; originally announced January 2025.

Improved residual mode separation for finite-dimensional control of PDEs: application to the Euler-Bernoulli beam

Authors: Anton Selivanov, Emilia Fridman

Abstract: We consider a simply-supported Euler-Bernoulli beam with viscous and Kelvin--Voigt damping. Our objective is to attenuate the effect of an unknown distributed disturbance using one piezoelectric actuator. We show how to design a suitable $H_\infty$ state-feedback controller based on a finite number of dominating modes. If the remaining (infinitely many) modes are ignored, the calculated $L^2$ gain… ▽ More We consider a simply-supported Euler-Bernoulli beam with viscous and Kelvin--Voigt damping. Our objective is to attenuate the effect of an unknown distributed disturbance using one piezoelectric actuator. We show how to design a suitable $H_\infty$ state-feedback controller based on a finite number of dominating modes. If the remaining (infinitely many) modes are ignored, the calculated $L^2$ gain is wrong. This happens because of the spillover phenomenon that occurs when the effect of the control on truncated modes is not accounted for in the feedback design. We propose a simple modification of the $H_\infty$ cost that prevents spillover. The key idea is to treat the control as a disturbance in the truncated modes and find the corresponding $L^2$ gains using the bounded real lemma. These $L^2$ gains are added to the control weight in the $H_\infty$ cost for the dominating modes, which prevents spillover. A numerical simulation of an aluminum beam with realistic parameters demonstrates the effectiveness of the proposed method. The presented approach is applicable to other types of PDEs, such as the heat, wave, and Kuramoto-Sivashinsky equations, as well as their semilinear versions. While this work focuses on $H_\infty$ control, the same methodology can be applied to guaranteed cost control, regional stability analysis, input-to-state stability, and systems with time-varying delays, including sampled-data systems. △ Less

Submitted 19 February, 2025; v1 submitted 10 August, 2023; originally announced August 2023.

Comments: This paper was accepted for publication in Systems and Control Letters. The preliminary version of the paper was accepted for presentation at IEEE Conference on Decision and Control 2023

Medical Image Captioning via Generative Pretrained Transformers

Authors: Alexander Selivanov, Oleg Y. Rogov, Daniil Chesakov, Artem Shelmanov, Irina Fedulova, Dmitry V. Dylov

Abstract: The automatic clinical caption generation problem is referred to as proposed model combining the analysis of frontal chest X-Ray scans with structured patient information from the radiology records. We combine two language models, the Show-Attend-Tell and the GPT-3, to generate comprehensive and descriptive radiology records. The proposed combination of these models generates a textual summary wit… ▽ More The automatic clinical caption generation problem is referred to as proposed model combining the analysis of frontal chest X-Ray scans with structured patient information from the radiology records. We combine two language models, the Show-Attend-Tell and the GPT-3, to generate comprehensive and descriptive radiology records. The proposed combination of these models generates a textual summary with the essential information about pathologies found, their location, and the 2D heatmaps localizing each pathology on the original X-Ray scans. The proposed model is tested on two medical datasets, the Open-I, MIMIC-CXR, and the general-purpose MS-COCO. The results measured with the natural language assessment metrics prove their efficient applicability to the chest X-Ray image captioning. △ Less

Submitted 28 September, 2022; originally announced September 2022.

Comments: 13 pages, 3 figures, The work was completed in 2021

An analysis of full-size Russian complexly NER labelled corpus of Internet user reviews on the drugs based on deep learning and language neural nets

Authors: Alexander Sboev, Sanna Sboeva, Ivan Moloshnikov, Artem Gryaznov, Roman Rybka, Alexander Naumov, Anton Selivanov, Gleb Rylkov, Viacheslav Ilyin

Abstract: We present the full-size Russian complexly NER-labeled corpus of Internet user reviews, along with an evaluation of accuracy levels reached on this corpus by a set of advanced deep learning neural networks to extract the pharmacologically meaningful entities from Russian texts. The corpus annotation includes mentions of the following entities: Medication (33005 mentions), Adverse Drug Reaction (17… ▽ More We present the full-size Russian complexly NER-labeled corpus of Internet user reviews, along with an evaluation of accuracy levels reached on this corpus by a set of advanced deep learning neural networks to extract the pharmacologically meaningful entities from Russian texts. The corpus annotation includes mentions of the following entities: Medication (33005 mentions), Adverse Drug Reaction (1778), Disease (17403), and Note (4490). Two of them - Medication and Disease - comprise a set of attributes. A part of the corpus has the coreference annotation with 1560 coreference chains in 300 documents. Special multi-label model based on a language model and the set of features is developed, appropriate for presented corpus labeling. The influence of the choice of different modifications of the models: word vector representations, types of language models pre-trained for Russian, text normalization styles, and other preliminary processing are analyzed. The sufficient size of our corpus allows to study the effects of particularities of corpus labeling and balancing entities in the corpus. As a result, the state of the art for the pharmacological entity extraction problem for Russian is established on a full-size labeled corpus. In case of the adverse drug reaction (ADR) recognition, it is 61.1 by the F1-exact metric that, as our analysis shows, is on par with the accuracy level for other language corpora with similar characteristics and the ADR representativnes. The evaluated baseline precision of coreference relation extraction on the corpus is 71, that is higher the results reached on other Russian corpora. △ Less

Submitted 30 April, 2021; originally announced May 2021.

Dynamic Traffic Reconstruction using Probe Vehicles

Authors: Matthieu Barreau, Anton Selivanov, Karl Henrik Johansson

Abstract: This article deals with the observation problem in traffic flow theory. The model used is the semilinear viscous Burgers equation. Instead of using the traditional fixed sensors to estimate the state of the traffic at given points, the measurements here are obtained from Probe Vehicles (PVs). We propose then a moving dynamic boundary observer whose boundaries are defined by the trajectories of the… ▽ More This article deals with the observation problem in traffic flow theory. The model used is the semilinear viscous Burgers equation. Instead of using the traditional fixed sensors to estimate the state of the traffic at given points, the measurements here are obtained from Probe Vehicles (PVs). We propose then a moving dynamic boundary observer whose boundaries are defined by the trajectories of the PVs. The main result of this article is the exponential convergence of the observation error, and, in some cases, its finite-time convergence. Finally, numerical simulations show that it is possible to observe the traffic in the congested, free-flow, and mixed regimes provided that the number of PVs is large enough. △ Less

Submitted 20 July, 2020; originally announced July 2020.

Comments: 6 pages, 3 figures, CDC conference paper

Multi-agent deployment under the leader displacement measurement: a PDE-based approach

Authors: Jieqiang Wei, Emilia Fridman, Anton Selivanov, Karl H. Johansson

Abstract: We study the deployment of a first-order multi-agent system over a desired smooth curve in 3D space. We assume that the agents have access to the local information of the desired curve and their displacements with respect to their closest neighbors, whereas in addition a leader is able to measure his absolute displacement with respect to the desired curve. In this paper we consider the case that t… ▽ More We study the deployment of a first-order multi-agent system over a desired smooth curve in 3D space. We assume that the agents have access to the local information of the desired curve and their displacements with respect to their closest neighbors, whereas in addition a leader is able to measure his absolute displacement with respect to the desired curve. In this paper we consider the case that the desired curve is a closed C^2 curve and we assume that the leader transmit his measurement to other agents through a communication network. We start the algorithm with displacement-based formation control protocol. Connections from this ODE model to a PDE model (heat equation), which can be seen as a reduced model, are then established. The resulting closed-loop system is modeled as a heat equation with delay (due to the communication). The boundary condition is periodic since the desired curve is closed. By choosing appropriate controller gains (the diffusion coefficient and the gain multiplying the leader state), we can achieve any desired decay rate provided the delay is small enough. The advantage of our approach is in the simplicity of the control law and the conditions. Numerical example illustrates the efficiency of the method. △ Less

Submitted 21 March, 2019; originally announced March 2019.

Sampled-data implementation of derivative-dependent control using artificial delays

Authors: Anton Selivanov, Emilia Fridman

Abstract: We study a sampled-data implementation of linear controllers that depend on the output and its derivatives. First, we consider an LTI system of relative degree $r\ge 2$ that can be stabilized using $r-1$ output derivatives. Then, we consider PID control of a second order system. In both cases, the Euler approximation is used for the derivatives giving rise to a delayed sampled-data controller. Giv… ▽ More We study a sampled-data implementation of linear controllers that depend on the output and its derivatives. First, we consider an LTI system of relative degree $r\ge 2$ that can be stabilized using $r-1$ output derivatives. Then, we consider PID control of a second order system. In both cases, the Euler approximation is used for the derivatives giving rise to a delayed sampled-data controller. Given a derivative-dependent controller that stabilizes the system, we show how to choose the parameters of the delayed sampled-data controller that preserves the stability under fast enough sampling. The maximum sampling period is obtained from LMIs that are derived using the Taylor's expansion of the delayed terms with the remainders compensated by appropriate Lyapunov-Krasovskii functionals. Finally, we introduce the event-triggering mechanism that may reduce the amount of sampled control signals used for stabilization. △ Less

Submitted 17 January, 2018; originally announced January 2018.

Predictor-based networked control under uncertain transmission delays

Authors: A. Selivanov, E. Fridman

Abstract: We consider state-feedback predictor-based control of networked control systems with large time-varying communication delays. We show that even a small controller-to-actuators delay uncertainty may lead to a non-small residual error in a networked control system and reveal how to analyze such systems. Then we design an event-triggered predictor-based controller with sampled measurements and demons… ▽ More We consider state-feedback predictor-based control of networked control systems with large time-varying communication delays. We show that even a small controller-to-actuators delay uncertainty may lead to a non-small residual error in a networked control system and reveal how to analyze such systems. Then we design an event-triggered predictor-based controller with sampled measurements and demonstrate that, depending on the delay uncertainty, one should choose various predictor models to reduce the error due to triggering. For the systems with a network only from a controller to actuators, we take advantage of the continuously available measurements by using a continuous-time predictor and employing a recently proposed switching approach to event-triggered control. By an example of an inverted pendulum on a cart we demonstrate that the proposed approach is extremely efficient when the uncertain time-varying network-induced delays are too large for the system to be stabilizable without a predictor. △ Less

Submitted 4 November, 2016; v1 submitted 17 November, 2015; originally announced November 2015.

Journal ref: Automatica, vol. 70, pp. 101-108 (2016)

Event-Triggered $H_\infty$ Control: a Switching Approach

Authors: Anton Selivanov, Emilia Fridman

Abstract: Event-triggered approach to networked control systems is used to reduce the workload of the communication network. For the static output-feedback continuous event-trigger may generate an infinite number of sampling instants in finite time (Zeno phenomenon) what makes it inapplicable to the real-world systems. Periodic event-trigger avoids this behavior but does not use all the available informatio… ▽ More Event-triggered approach to networked control systems is used to reduce the workload of the communication network. For the static output-feedback continuous event-trigger may generate an infinite number of sampling instants in finite time (Zeno phenomenon) what makes it inapplicable to the real-world systems. Periodic event-trigger avoids this behavior but does not use all the available information. In the present paper we aim to exploit the advantage of the continuous-time measurements and guarantee a positive lower bound on the inter-event times by introducing a switching approach for finding a waiting time in the event-triggered mechanism. Namely, our idea is to present the closed-loop system as a switching between the system under periodic sampling and the one under continuous event-trigger and take the maximum sampling preserving the stability as the waiting time. We extend this idea to the $L_2$-gain and ISS analysis of perturbed networked control systems with network-induced delays. By examples we demonstrate that the switching approach to event-triggered control can essentially reduce the amount of measurements to be sent through a communication network compared to the existing methods. △ Less

Submitted 4 November, 2016; v1 submitted 4 June, 2015; originally announced June 2015.

Comments: 14 pages, 4 figures

Journal ref: IEEE Transactions on Automatic Control, vol. 61, no. 10, 2016

Controlling cluster synchronization by adapting the topology

Authors: Judith Lehnert, Philipp Hövel, Anton Selivanov, Alexander Fradkov, Eckehard Schöll

Abstract: We suggest an adaptive control scheme for the control of zero-lag and cluster synchronization in delay-coupled networks. Based on the speed-gradient method, our scheme adapts the topology of a network such that the target state is realized. It is robust towards different initial condition as well as changes in the coupling parameters. The emerging topology is characterized by a delicate interplay… ▽ More We suggest an adaptive control scheme for the control of zero-lag and cluster synchronization in delay-coupled networks. Based on the speed-gradient method, our scheme adapts the topology of a network such that the target state is realized. It is robust towards different initial condition as well as changes in the coupling parameters. The emerging topology is characterized by a delicate interplay of excitatory and inhibitory links leading to the stabilization of the desired cluster state. As a crucial parameter determining this interplay we identify the delay time. Furthermore, we show how to construct networks such that they exhibit not only a given cluster state but also with a given oscillation frequency. We apply our method to coupled Stuart-Landau oscillators, a paradigmatic normal form that naturally arises in an expansion of systems close to a Hopf bifurcation. The successful and robust control of this generic model opens up possible applications in a wide range of systems in physics, chemistry, technology, and life science. △ Less

Submitted 15 August, 2014; v1 submitted 30 May, 2014; originally announced May 2014.

Journal ref: Phys. Rev. E 90, 4, 042914 (2014)

Adaptive synchronization in delay-coupled networks of Stuart-Landau oscillators

Authors: Anton Selivanov, Judith Lehnert, Thomas Dahms, Philipp Hövel, Alexander L. Fradkov, Eckehard Schöll

Abstract: We consider networks of delay-coupled Stuart-Landau oscillators. In these systems, the coupling phase has been found to be a crucial control parameter. By proper choice of this parameter one can switch between different synchronous oscillatory states of the network. Applying the speed-gradient method, we derive an adaptive algorithm for an automatic adjustment of the coupling phase such that a des… ▽ More We consider networks of delay-coupled Stuart-Landau oscillators. In these systems, the coupling phase has been found to be a crucial control parameter. By proper choice of this parameter one can switch between different synchronous oscillatory states of the network. Applying the speed-gradient method, we derive an adaptive algorithm for an automatic adjustment of the coupling phase such that a desired state can be selected from an otherwise multistable regime. We propose goal functions based on both the difference of the oscillators and a generalized order parameter and demonstrate that the speed-gradient method allows one to find appropriate coupling phases with which different states of synchronization, e.g., in-phase oscillation, splay or various cluster states, can be selected. △ Less

Submitted 21 July, 2011; originally announced July 2011.

Comments: 8 pages, 7 figures

Journal ref: Phys. Rev. E 85, 016201 (2012)

Composite S-Brane Solutions On Product Of Ricci-Flat Spaces

Authors: V. D. Ivashchuk, V. N. Melnikov, A. B. Selivanov

Abstract: A family of generalized $S$-brane solutions with orthogonal intersection rules and $n$ Ricci-flat factor spaces in the theory with several scalar fields and antisymmetric forms is considered. Two subclasses of solutions with power-law and exponential behaviour of scale factors are singled out. These subclasses contain sub-families of solutions with accelerated expansion of certain factor spaces.… ▽ More A family of generalized $S$-brane solutions with orthogonal intersection rules and $n$ Ricci-flat factor spaces in the theory with several scalar fields and antisymmetric forms is considered. Two subclasses of solutions with power-law and exponential behaviour of scale factors are singled out. These subclasses contain sub-families of solutions with accelerated expansion of certain factor spaces. The solutions depend on charge densities of branes, their dimensions and intersections, dilatonic couplings and the number of dilatonic fields. △ Less

Submitted 16 April, 2004; originally announced April 2004.

Comments: To appear in GRG

Journal ref: Gen.Rel.Grav. 36 (2004) 1593-1602

Cosmological solutions in multidimensional model with multiple exponential potential

Authors: V. D. Ivashchuk, V. N. Melnikov, A. B. Selivanov

Abstract: A family of cosmological solutions with $(n+1)$ Ricci-flat spaces in the theory with several scalar fields and multiple exponential potential is obtained when coupling vectors in exponents obey certain relations. Two subclasses of solutions with power-law and exponential behaviour of scale factors are singled out. It is proved that power-law solutions may take place only when coupling vectors ar… ▽ More A family of cosmological solutions with $(n+1)$ Ricci-flat spaces in the theory with several scalar fields and multiple exponential potential is obtained when coupling vectors in exponents obey certain relations. Two subclasses of solutions with power-law and exponential behaviour of scale factors are singled out. It is proved that power-law solutions may take place only when coupling vectors are linearly independent and exponential dependence occurs for linearly dependent set of coupling vectors. A subfamily of solutions with accelerated expansion is singled out. A generalized isotropization behaviours of certain classes of general solutions are found. In quantum case exact solutions to Wheeler-DeWitt equation are obtained and special "ground state" wave functions are considered. △ Less

Submitted 23 September, 2003; v1 submitted 18 August, 2003; originally announced August 2003.

Comments: 22 pages, 1 figure

Journal ref: JHEP 0309 (2003) 059

Simulation of intersecting black brane solutions

Authors: V. D. Ivashchuk, V. N. Melnikov, A. B. Selivanov

Abstract: A family of spherically symmetric solutions with horizon in the model with multi-component anisotropic fluid (MCAF) is obtained. The metric of any solution contains n-1 Ricci-flat ``internal space'' metrics and for certain equations of state (p_i = \pm ρ) coincides with the metric of intersecting black brane solution in the model with antisymmetric forms. Examples of simulation of intersecting M… ▽ More A family of spherically symmetric solutions with horizon in the model with multi-component anisotropic fluid (MCAF) is obtained. The metric of any solution contains n-1 Ricci-flat ``internal space'' metrics and for certain equations of state (p_i = \pm ρ) coincides with the metric of intersecting black brane solution in the model with antisymmetric forms. Examples of simulation of intersecting M2 and M5 black branes are considered. The post-Newtonian parameters beta and gamma corresponding to the 4-dimensional section of the metric are calculated. △ Less

Submitted 26 November, 2002; originally announced November 2002.

Comments: 7 pages, 1 figure

Journal ref: Grav.Cosmol. 9 (2003) 50-54

Multidimensional Black Hole Solutions In Model With Perfect Fluid

Authors: V. D. Ivashchuk, V. N. Melnikov, A. B. Selivanov

Abstract: A family of black-hole solutions in the model with 1-component perfect fluid is obtained. The metric of any solution contains (n -1) Ricci-flat "internal space" metrics and for certain equations of state coincides with the metric of black brane (or black hole) solution in the model with antisymmetric form. Certain examples (e.g. imitating M2 and M5 black branes) are considered. The post-Newtonia… ▽ More A family of black-hole solutions in the model with 1-component perfect fluid is obtained. The metric of any solution contains (n -1) Ricci-flat "internal space" metrics and for certain equations of state coincides with the metric of black brane (or black hole) solution in the model with antisymmetric form. Certain examples (e.g. imitating M2 and M5 black branes) are considered. The post-Newtonian parameters beta and gamma corresponding to the 4-dimensional section of the metric are calculated. △ Less

Submitted 23 May, 2002; originally announced May 2002.

Comments: 5 pages, LaTeX, published in Grav. Cosmol., 7, 4

Journal ref: Grav.Cosmol.7:308-310,2001

Linux-Based Toolkit in the VEPP-4 Control System

Authors: V. Blinov, A. Bogomyagkov, M. Fedotov, S. Karnaev, O. Meshkov, N. Muchnoi, S. Nikitin, I. Nikolaev, A. Selivanov, S. Smirnov, V. Tsukanov

Abstract: The control system of the VEPP-4 facility was designed more than fifteen years ago and based on the home-developed CAMAC-embedded minicomputers Odrenok [1]. Five years ago, all computers were connected via Ethernet network. This step allowed us to force an integration of PCs into the VEPP-4 control system. This paper reviews new tools running on the PCs under Linux. The beam diagnostics tool des… ▽ More The control system of the VEPP-4 facility was designed more than fifteen years ago and based on the home-developed CAMAC-embedded minicomputers Odrenok [1]. Five years ago, all computers were connected via Ethernet network. This step allowed us to force an integration of PCs into the VEPP-4 control system. This paper reviews new tools running on the PCs under Linux. The beam diagnostics tool described in this paper provides the data about position and dimensions of e-/e+ beams from CCD matrix controller via 100-Mbit Ethernet. The next tool provides measuring of beam energy at the VEPP-4M collider using a well-known method of resonance depolarization by the observation of the polarization degree on the effect of internal scattering of particles. Data visualization tools are based on CERN ROOT framework. Hardware and software aspects of the systems are presented in this paper. △ Less

Submitted 17 November, 2001; originally announced November 2001.

Comments: 3 pages, 4 figures, ICALEPCS'01, PSN#WEAP028

Journal ref: eConf C011127:WEAP028,2001