|
«INFORMATICS AND APPLICATIONS»
Scientific journal
Volume 20, Issue 2, 2026
Content | About Authors
Abstract and Keywords
- I. V. Chizhov M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Abstract: The Schur-Hadamard square of a random subcode of a random linear code over a finite field of characteristic 2 is studied. Namely, for a uniformly random generator (k x n)-matrix G of a code C and a uniformly
It is proved 
the above probability differs from one by a quantity that is exponentially small in n, and this holds simultaneously for an overwhelming fraction of matrices G. This result provides a rigorous justification of an experimentally observed property that underlies several attacks on McEliece-type code-based cryptosystems built upon subcodes of generalized Reed-Solomon codes and algebraic-geometric codes. As a technical tool, exact formulas and tight upper bounds are derived for the number of totally isotropic subspaces of a given dimension of an arbitrary symmetric bilinear form of prescribed rank over a field of characteristic 2; these bounds maybe of independent interest.
Keywords: Schur square; random subcode of a linear code; totally isotropic subspace of a symmetric bilinear form; code-based cryptography
- D. V. Vinogradov Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Abstract: The paper presents a polynomial-time algorithm (with respect to the size of the initial training sample) for constructing the smallest training set for a factor-lattice of the lattice of candidates for the smallest training sample through a sequence of reductions of this sample. The structure of the smallest training sample is described alongside a proof of its minimality. The proposed algorithm can be useful for consistently reducing the space of features describing training examples so as to preserve the relationships between the elements of the lattice as much as possible. Then, if necessary, it will be possible to return to the original representation and explore the local neighborhood of the element of interest. The construction is based on the well-known characterization of lattice congruences, a simple proof of this fact is also given in the article. The correctness and completeness theorems for the algorithm are proved. The examples are given to demonstrate the subtleties of the application and operation of the algorithm presented in the work.
Keywords: lattice; irreducible elements; training sample; candidate; congruence; factor-lattice
- D. D. Dorin Moscow Institute of Physics and Technology (National Research University), 9 Institutskiy Per., Dolgoprudny, Moscow Region 141701, Russian Federation
- N. S. Kiselev Moscow Institute of Physics and Technology (National Research University), 9 Institutskiy Per., Dolgoprudny, Moscow Region 141701, Russian Federation
- A. V. Grabovoy Moscow Institute of Physics and Technology (National Research University), 9 Institutskiy Per., Dolgoprudny, Moscow Region 141701, Russian Federation, V.A. Trapeznikov Institute of Control Sciences of the Russian Academy of Sciences, 65 Profsoyuznaya Str., Moscow 117997, Russian Federation
Abstract: Reconstructing visual stimuli from neural signals is a fundamental challenge in neurodecoding, lying at the intersection of computational neuroscience and modern machine learning. Despite recent advances achieved using contrastive representations and diffusion-based generative models, most existing approaches are limited to a single neuroimaging modality - either functional magnetic resonance imaging (fMRI) with high spatial resolution, or electroencephalography (EEG) with high temporal resolution. The integration of both modalities remains a largely unexplored area. In this work, a multimodal architecture is proposed that jointly processes fMRI and EEG signals to reconstruct visual stimuli. Brain activity embeddings are trained contrastively to align with CLIP image embeddings. The proposed two-stage generation pipeline comprises synthesis of an intermediate latent representation via a prior model trained onjoint fMRI-EEG vectors, followed by decoding this representation using a pretrained diffusion model conditioned on CLIP embeddings. Experiments on a publicly available multimodal dataset demonstrate the efficacy of the proposed architecture for neural decoding. Quantitatively, the multimodal model surpasses unimodal baselines in terms of CLIP-Score, underscoring the importance of jointly leveraging fMRI and EEG signals for accurate visual stimulus reconstruction.
Keywords: visual stimulus decoding; fMRI-EEG; contrastive learning; diffusion models; image reconstruction
- A. V. Panteleev Moscow Aviation Institute (National Research University), 4 Volokolamskoe Shosse, Moscow 125933, Russian Federation
- E. A. Khvoshnyanskaya Moscow Aviation Institute (National Research University), 4 Volokolamskoe Shosse, Moscow 125933, Russian Federation
Abstract: The paper is devoted to the development and application of metaheuristic optimization algorithms to solve optimal control problems for continuous and discrete dynamic systems. A search strategy and an algorithm for finding the extremum of a multivariable function under interval constraints are described. The search procedure leverages concepts from metaheuristic optimization algorithms regarding dynamic modifications of the search space, alongside accelerated algorithms based on prediction and the incorporation of successful solution memory obtained during the exploration of the feasible solution set. The efficiency of the proposed method is demonstrated through examples of solving optimal open-loop control problems forboth continuous and discrete dynamic systems.
In the latter case, the control problems for an individual trajectory, a bundle of trajectories under initial condition uncertainties, and a family of trajectories of a stochastic system are considered. The problems of parametric optimization of technical systems, including a tension/compression spring, a three-bar truss, and a tubular column, are solved. The presented numerical results are accompanied by recommendations for tuning the hyperparameters of the proposed optimization method.
Keywords: metaheuristic optimization algorithms; optimal control; open-loop control; dynamic systems
- A. V. Borisov Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation, M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation
- Yu. N. Kurinov M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation
Abstract: The paper continues a series of studies devoted to the analysis and estimation problems for a class of special Markov jump processes. It addresses the filtering problem for special Markov jump processes based on discretized observations represented by increments of a diffusion process whose drift and diffusion coefficients depend on the state of the signal process. The objective is to determine the conditional distribution of the estimated signal with respect to the available observations. The equations of optimal filtering are derived and a numerical algorithm for their implementation is proposed based on constructing analytical approximations of the corresponding conditional densities. A statement characterizing the approximation accuracy as a function of the approximation order is proven. The performance of the proposed estimates is illustrated by a numerical example.
Keywords: special Markovjump process; discretized observations; observation with multiplicative noise; conditional probability density; analytical approximation of filtering estimate
- A. A. Kudryavtsev Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Moscow Center for Fundamental and Applied Mathematics, M.V. Lomonosov Moscow State University, 1 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation
- O. V. Shestakov Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Moscow Center for Fundamental and Applied Mathematics, M.V. Lomonosov Moscow State University, 1 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Abstract: The paper considers the digamma distribution, special cases of which can be represented as a generalized gamma distribution and a generalized beta distribution of the second kind. The strong consistency and asymptotic normality of digamma distribution parameter estimates obtained using a modified method of moments based on sample cumulants are proved in the case of weakly dependent sample components. Estimates of three unknown parameters (the characteristic exponent, shape parameters, and scale parameters) are considered for fixed concentration parameters. Estimation of the latter is not considered due to the nontrivial nature of the inversion of polygamma functions. The formulated statements pertain to a limited set of values of the estimated parameters; however, they can be easily extended to the general case using an estimation algorithm detailed in the authors' previous cited papers. The proof of the main statement is based on a sufficient condition for weak convergence of functions of asymptotically normal vectors. The results of this paper can be used to describe a wide class of distributions arising in the description of processes modeled by distributions with nonnegative, unbounded support.
Keywords: weak dependence; parameter estimation; digamma distribution; mixed distributions; method of moments; cumulants; asymptotic normality
- A. A. Grusho Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
- N. A. Grusho Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
- M. I. Zabezhailo Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
- A. A. Zatsarinny Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
- V. O. Piskovski Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Abstract: The use of network-centric technologies in control and decision support systems is determined by the concept of development of control and monitoring of peripheral nodes of the network-centric system (SCS) and the conditions for ensuring information security. Security issues in SCS have a number of specific features. The main goal of the Center can be described as a controlling cause consisting of properties that ultimately generate actions throughout the system. The control cause consists of a plurality of separate properties which themselves serve as the causes for sets of properties of slave SCS nodes functioning simultaneously as derived causes. The paper discusses problems of monitoring anomalies in the SCS when performing the main task set by the Center for peripheral nodes. Three types of anomalies were identified: (i) distortion of the properties of the consequences of derivative causes that give rise to actions to implement the plans of the Center; (ii) presence of obstacles when performing actions by nodes in the interests of the Center; and (iii) increased resource consumption which may slow down or stop the execution of actions at the remote node in the interests of the Center. The greatest difficulties are associated with the identification of anomalies of the first type.
Keywords: information security; network-centric systems; anomalies; detection of anomalies
- A. K. Bergovin Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Moscow Center for Fundamental and Applied Mathematics, M.V. Lomonosov Moscow State University, 1 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation
- V. G. Ushakov Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V Lomonosov Moscow State University, 1-52 Leninskie Gory, GSP-1, Moscow 119991, Russian Federation, Federal Research Center "Computer Science and Control" of the Russian Academy of Sciences, 44-2 Vavilov Str., Moscow 119333, Russian Federation
Abstract: A single-server queueing system with an unlimited waiting capacity and arbitrary service time distribution is studied, in which the arrival rate of the Poisson input process depends on the last event in the system: an arrival or a service completion. Input flows of this structure make it possible to model situations where the behavior of the incoming flow depends on the system's operation. The method of supplementary components serves as the mathematical apparatus of the study, with the help of which the distribution of the number of customers in the system was found in the nonstationary regime and, as a consequence, in the stationary regime as well. Given that most real-world systems are heavily loaded, there arises a need to investigate the system's characteristics under critical load conditions; therefore, the second part of the paper is devoted to this task. Based on the analysis of the queue-length behavior under critical load, the limiting distribution of the number of customers in the system is obtained in explicit form.
Keywords: Poisson flow; event-dependent rate; queue length; supplementary variable method; heavy traffic
|
