Sergey V. Samsonov
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Sergey V. Samsonov has a total of 247 co-authors. Here are the top co-authors by number of shared publications:
- Vladimir L. Bratman
Ariel University
50 shared publications G. G. Denisov
A. V. Gaponov-Grekhov Institute of Applied Physics of the Russian Academy of Sciences, Nizhny Novgorod, Russia
30 shared publications- Alan D. R. Phelps
Department of Physics, SUPA, University of Strathclyde, Glasgow, U.K.
28 shared publications - Adrian W. Cross
University of Strathclyde, Glasgow, United Kingdom
23 shared publications K. Ronald
Department of Physics, SUPA, University of Strathclyde, 107 Rottenrow, Glasgow, G4 0NG, Scotland, UK
21 shared publicationsW. He
Department of Physics, SUPA, University of Strathclyde, 107 Rottenrow, Glasgow, G4 0NG, Scotland, UK
21 shared publications
Latest Publications
- 16 August 2026
- Published by Taylor & Francis in Statistics
- p. 1-25
On the rate of Gaussian approximation for online linear regression problems
1. Gaussian approximation results play a central role in modern machine learning, where algorithms are often implemented via stochastic approximation methods such as stochastic gradient descent (SG...
- 16 August 2026
- Published by Taylor & Francis in Statistics
- p. 1-25
On the rate of Gaussian approximation for online linear regression problems
1. Gaussian approximation results play a central role in modern machine learning, where algorithms are often implemented via stochastic approximation methods such as stochastic gradient descent (SG...
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 1310-1321
Bradley-Terry Rankings for Recommender Systems Across Dataset Taxonomies
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 1310-1321
Open AccessBradley-Terry Rankings for Recommender Systems Across Dataset Taxonomies
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 8637-8640
Revisiting IPS-based Algorithms for Off-Policy Evaluation of Contextual Bandits
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 8637-8640
Open AccessRevisiting IPS-based Algorithms for Off-Policy Evaluation of Contextual Bandits
- 31 March 2026
- Published by LLC Science and Innovations in Psychosomatic and Integrative Research
- Vol. 12 (1)
Psychotropic effects of space weather factors during the period of maximum solar activity in volunteers of the "Heliomed 2" project living in middle latitudes
Despite a significant number of publications devoted to the biotropic effects of space weather factors, the psychotropic effects of heliogeophysical factors remain largely unexplored. Objective. To determine the psychological characteristics of female participants in the "Heliomed – 2" project living in mid-latitudes on one of its stages during the period of maximum solar activity, taking into account the psychological heliogeomagnetic synchronization identified in them. Materials and Methods. To achieve this objective, 13 female volunteers with an average age of 41.2 years (40.2–44.6) living in mid-latitudes were surveyed during the "Heliomed – 2" project in 2024 (a year of high-solar activity) using the C. Spielberger – J. Hanin questionnaire, E. Haim’s stress-coping behavior test, and S. Dellinger’s projective psychogeometric test; in addition, the integral indicator of heliogeomagnetic disturbance, the Kp-index, was determined. When comparing the daily measured indicators of reactive anxiety and the Kp-index, all subjects were divided into two groups: Group I – individuals possessing psychological heliogeomagnetic synchronization; Group II – individuals with no such synchronization. Results. It was found that at this stage, individuals in Group II (without heliogeophysical synchronization) predominated; compared to volunteers in Group I, they were more anxious, less effective in developing coping behaviors (primarily in the emotional and behavioral spheres), actively suppressed their emotions, engaged in self-blame, and sought to avoid emerging problems, therefore, they could be classified as a high-risk group for the development of psychosomatic disorders and cardiovascular events. Conclusion. Thus, when compared with the results of previous studies of volunteers at different latitudes and during different periods of solar activity, further research must take into account gender, latitude of residence, and solar cycle activity.
- 31 March 2026
- Published by LLC Science and Innovations in Psychosomatic and Integrative Research
- Vol. 12 (1)
Psychotropic effects of space weather factors during the period of maximum solar activity in volunteers of the "Heliomed 2" project living in middle latitudes
Despite a significant number of publications devoted to the biotropic effects of space weather factors, the psychotropic effects of heliogeophysical factors remain largely unexplored. Objective. To determine the psychological characteristics of female participants in the "Heliomed – 2" project living in mid-latitudes on one of its stages during the period of maximum solar activity, taking into account the psychological heliogeomagnetic synchronization identified in them. Materials and Methods. To achieve this objective, 13 female volunteers with an average age of 41.2 years (40.2–44.6) living in mid-latitudes were surveyed during the "Heliomed – 2" project in 2024 (a year of high-solar activity) using the C. Spielberger – J. Hanin questionnaire, E. Haim’s stress-coping behavior test, and S. Dellinger’s projective psychogeometric test; in addition, the integral indicator of heliogeomagnetic disturbance, the Kp-index, was determined. When comparing the daily measured indicators of reactive anxiety and the Kp-index, all subjects were divided into two groups: Group I – individuals possessing psychological heliogeomagnetic synchronization; Group II – individuals with no such synchronization. Results. It was found that at this stage, individuals in Group II (without heliogeophysical synchronization) predominated; compared to volunteers in Group I, they were more anxious, less effective in developing coping behaviors (primarily in the emotional and behavioral spheres), actively suppressed their emotions, engaged in self-blame, and sought to avoid emerging problems, therefore, they could be classified as a high-risk group for the development of psychosomatic disorders and cardiovascular events. Conclusion. Thus, when compared with the results of previous studies of volunteers at different latitudes and during different periods of solar activity, further research must take into account gender, latitude of residence, and solar cycle activity.
- 3 February 2026
- Published by Springer Nature in Journal of Optimization Theory and Applications
- Vol. 208 (3) , 89
UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms
- 3 February 2026
- Published by Springer Nature in Journal of Optimization Theory and Applications
- Vol. 208 (3) , 89
UVIP: Model-Free Approach to Evaluate Reinforcement Learning Algorithms
- 1 December 2025
- Published by Kotelnikov Institute of Radioengineering and Electronics of Russian Academy of Sciences in Journal of Radio Electronics
- Vol. 2025 (12)
Experimental observation of self-mode locking regimes in a gyro-resonant backward-wave oscillator with a feedback circuit
A pulse-periodic oscillation regime was experimentally obtained in a gyro-resonant backward-wave oscillator (BWO) with a feedback circuit providing a high (approximately 0.9) transmission coefficient. The experiments were conducted by a 30 GHz gyro-BWO with a working space in the form of a three-pass corrugated waveguide. The pulse repetition frequency, with duration of ~1.5 ns, was approximately 220 MHz. The observed regime is interpreted as a mode-locking effect in a ring resonator formed by the interaction space of the gyro-BWO and the feedback circuit.
- 1 December 2025
- Published by Kotelnikov Institute of Radioengineering and Electronics of Russian Academy of Sciences in Journal of Radio Electronics
- Vol. 2025 (12)
Experimental observation of self-mode locking regimes in a gyro-resonant backward-wave oscillator with a feedback circuit
A pulse-periodic oscillation regime was experimentally obtained in a gyro-resonant backward-wave oscillator (BWO) with a feedback circuit providing a high (approximately 0.9) transmission coefficient. The experiments were conducted by a 30 GHz gyro-BWO with a working space in the form of a three-pass corrugated waveguide. The pulse repetition frequency, with duration of ~1.5 ns, was approximately 220 MHz. The observed regime is interpreted as a mode-locking effect in a ring resonator formed by the interaction space of the gyro-BWO and the feedback circuit.
- Open Access
gfnx: Fast and Scalable Library for Generative Flow Networks in JAX
In this paper, we present gfnx, a fast and scalable package for training and evaluating Generative Flow Networks (GFlowNets) written in JAX. gfnx provides an extensive set of environments and metrics for benchmarking, accompanied with single-file implementations of core objectives for training GFlowNets. We include synthetic hypergrids, multiple sequence generation environments with various editing regimes and particular reward designs for molecular generation, phylogenetic tree construction, Bayesian structure learning, and sampling from the Ising model energy. Across different tasks, gfnx achieves significant wall-clock speedups compared to Pytorch-based benchmarks (such as torchgfn library) and author implementations. For example, gfnx achieves up to 55 times speedup on CPU-based sequence generation environments, and up to 80 times speedup with the GPU-based Bayesian network structure learning setup. Our package provides a diverse set of benchmarks and aims to standardize empirical evaluation and accelerate research and applications of GFlowNets. The library is available on GitHub (https://github.com/d-tiapkin/gfnx) and on pypi (https://pypi.org/project/gfnx/). Documentation is available on https://gfnx.readthedocs.io.
Open Accessgfnx: Fast and Scalable Library for Generative Flow Networks in JAX
In this paper, we present gfnx, a fast and scalable package for training and evaluating Generative Flow Networks (GFlowNets) written in JAX. gfnx provides an extensive set of environments and metrics for benchmarking, accompanied with single-file implementations of core objectives for training GFlowNets. We include synthetic hypergrids, multiple sequence generation environments with various editing regimes and particular reward designs for molecular generation, phylogenetic tree construction, Bayesian structure learning, and sampling from the Ising model energy. Across different tasks, gfnx achieves significant wall-clock speedups compared to Pytorch-based benchmarks (such as torchgfn library) and author implementations. For example, gfnx achieves up to 55 times speedup on CPU-based sequence generation environments, and up to 80 times speedup with the GPU-based Bayesian network structure learning setup. Our package provides a diverse set of benchmarks and aims to standardize empirical evaluation and accelerate research and applications of GFlowNets. The library is available on GitHub (https://github.com/d-tiapkin/gfnx) and on pypi (https://pypi.org/project/gfnx/). Documentation is available on https://gfnx.readthedocs.io.
- Open Access
Improved Central Limit Theorem and Bootstrap Approximations for Linear Stochastic Approximation
In this paper, we refine the Berry-Esseen bounds for the multivariate normal approximation of Polyak-Ruppert averaged iterates arising from the linear stochastic approximation (LSA) algorithm with decreasing step size. We consider the normal approximation by the Gaussian distribution with covariance matrix predicted by the Polyak-Juditsky central limit theorem and establish the rate up to order $n^{-1/3}$ in convex distance, where $n$ is the number of samples used in the algorithm. We also prove a non-asymptotic validity of the multiplier bootstrap procedure for approximating the distribution of the rescaled error of the averaged LSA estimator. We establish approximation rates of order up to $1/\sqrt{n}$ for the latter distribution, which significantly improves upon the previous results obtained by Samsonov et al. (2024).Open AccessImproved Central Limit Theorem and Bootstrap Approximations for Linear Stochastic Approximation
In this paper, we refine the Berry-Esseen bounds for the multivariate normal approximation of Polyak-Ruppert averaged iterates arising from the linear stochastic approximation (LSA) algorithm with decreasing step size. We consider the normal approximation by the Gaussian distribution with covariance matrix predicted by the Polyak-Juditsky central limit theorem and establish the rate up to order $n^{-1/3}$ in convex distance, where $n$ is the number of samples used in the algorithm. We also prove a non-asymptotic validity of the multiplier bootstrap procedure for approximating the distribution of the rescaled error of the averaged LSA estimator. We establish approximation rates of order up to $1/\sqrt{n}$ for the latter distribution, which significantly improves upon the previous results obtained by Samsonov et al. (2024). - 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 46 (11) , 2153-2156
Ka-Band Inverted Gyrotron With 80% Efficiency of Microwave-to-DC Power Conversion
The paper presents the results of the first experiments on the investigation of an inverted gyrotron or a gyrotron rectifier, a device which, unlike a conventional gyrotron, converts incident microwave power into direct current (DC) power. In the experiment, by means of cyclotron-resonance interaction with a thin solid initially rectilinear 8.4 keV/0.54 A electron beam, microwave radiation at a frequency of about 34 GHz with a power of 13 kW is converted by it into a power of 15 kW of a 70-μs pulse current that flows through a 51.4 kΩ resistor connecting the device collector to ground. The achieved efficiency of microwave-to-DC power conversion, taking into account the power to drive the electron beam, is 80%.
- 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Electron Device Letters
- Vol. 46 (11) , 2153-2156
Ka-Band Inverted Gyrotron With 80% Efficiency of Microwave-to-DC Power Conversion
The paper presents the results of the first experiments on the investigation of an inverted gyrotron or a gyrotron rectifier, a device which, unlike a conventional gyrotron, converts incident microwave power into direct current (DC) power. In the experiment, by means of cyclotron-resonance interaction with a thin solid initially rectilinear 8.4 keV/0.54 A electron beam, microwave radiation at a frequency of about 34 GHz with a power of 13 kW is converted by it into a power of 15 kW of a 70-μs pulse current that flows through a 51.4 kΩ resistor connecting the device collector to ground. The achieved efficiency of microwave-to-DC power conversion, taking into account the power to drive the electron beam, is 80%.
- Open Access
On the Rate of Gaussian Approximation for Linear Regression Problems
In this paper, we consider the problem of Gaussian approximation for the online linear regression task. We derive the corresponding rates for the setting of a constant learning rate and study the explicit dependence of the convergence rate upon the problem dimension $d$ and quantities related to the design matrix. When the number of iterations $n$ is known in advance, our results yield the rate of normal approximation of order $\sqrt{\log{n}/n}$, provided that the sample size $n$ is large enough.Open AccessOn the Rate of Gaussian Approximation for Linear Regression Problems
In this paper, we consider the problem of Gaussian approximation for the online linear regression task. We derive the corresponding rates for the setting of a constant learning rate and study the explicit dependence of the convergence rate upon the problem dimension $d$ and quantities related to the design matrix. When the number of iterations $n$ is known in advance, our results yield the rate of normal approximation of order $\sqrt{\log{n}/n}$, provided that the sample size $n$ is large enough.