All issues

This paper presents the results of experimental validation of some structural issues concerning the practical use of methods to overcome catastrophic forgetting of neural networks. A comparison of current effective methods like EWC (Elastic Weight Consolidation) and WVA (Weight Velocity Attenuation) is made and their advantages and disadvantages are considered. It is shown that EWC is better for tasks where full retention of learned skills is required on all the tasks in the training queue, while WVA is more suitable for sequential tasks with very limited computational resources, or when reuse of representations and acceleration of learning from task to task is required rather than exact retention of the skills. The attenuation of the WVA method must be applied to the optimization step, i. e. to the increments of neural network weights, rather than to the loss function gradient itself, and this is true for any gradient optimization method except the simplest stochastic gradient descent (SGD). The choice of the optimal weights attenuation function between the hyperbolic function and the exponent is considered. It is shown that hyperbolic attenuation is preferable because, despite comparable quality at optimal values of the hyperparameter of the WVA method, it is more robust to hyperparameter deviations from the optimal value (this hyperparameter in the WVA method provides a balance between preservation of old skills and learning a new skill). Empirical observations are presented that support the hypothesis that the optimal value of this hyperparameter does not depend on the number of tasks in the sequential learning queue. And, consequently, this hyperparameter can be picked up on a small number of tasks and used on longer sequences.

One of the destroying natural processes is the initiation of the regional seismic activity. It leads to a large number of human deaths. Much effort has been made to develop precise and robust methods for the estimation of the seismic stability of buildings. One of the most common approaches is the natural frequency method. The obvious drawback of this approach is a low precision due to the model oversimplification. The other method is a detailed simulation of dynamic processes using the finite-element method. Unfortunately, the quality of simulations is not enough due to the difficulty of setting the correct free boundary condition. That is why the development of new numerical methods for seismic stability problems is a high priority nowadays.

The present work is devoted to the study of spatial dynamic processes occurring in geological medium during an earthquake. We describe a method for simulating seismic wave propagation from the hypocenter to the day surface. To describe physical processes, we use a system of partial differential equations for a linearly elastic body of the second order, which is solved numerically by a grid-characteristic method on parallelepiped meshes. The widely used geological hypocenter model, called the “double-couple” model, was incorporated into this numerical algorithm. In this case, any heterogeneities, such as geological layers with curvilinear boundaries, gas and fluid-filled cracks, fault planes, etc., may be explicitly taken into account.

In this paper, seismic waves emitted during the earthquake initiation process are numerically simulated. Two different models are used: the homogeneous half-space and the multilayered geological massif with the day surface. All of their parameters are set based on previously published scientific articles. The adequate coincidence of the simulation results is obtained. And discrepancies may be explained by differences in numerical methods used. The numerical approach described can be extended to more complex physical models of geological media.

The accurate characterization of particle size and morphology is crucial across numerous scientific and industrial fields. While microscopy is a powerful tool for visualizing particle geometry, traditional image analysis often relies on basic parameters like area and equivalent circular diameter, which are insufficient for describing non-spherical or complex particles. This paper presents a novel methodology for the analysis of particles from micrographs, combining deep learningbased segmentation with parametric polygon approximation. Optical microscopy images were obtained for model systems of increasing complexity: monodisperse polystyrene spheres, SAPO-34 cubes, ZSM-5 hexagonal prisms, and a multi-component mixture of all three. Neural network segmentation was performed using the DLgram01 cloud service, trained on manually annotated images. A custom Python program was developed to process the resulting segmentation polygons. The core of the method involves fitting a parameterized polygon to each segmented particle’s contour. The optimal value of geometric parameters for each particle is determined by minimizing a “dissimilarity” function that measures the discrepancy between the segmentation and the parametric shape. This approach was validated on the simple spherical system, yielding a size distribution consistent with conventional projected diameter methods. For more complex cubic SAPO-34 particles, the rectangle fit revealed deviations from the ideal shape, providing a more complete morphological description than using a single size parameter. In the mixture of particles, the method successfully classified and extracted distinct parameter distributions for each particle type. The results for each class in the mixture were statistically consistent with those obtained from the single-component systems, demonstrating the method’s robustness and accuracy. The proposed methodology provides a powerful and universal tool for automated, high-throughput particle analysis, enabling the extraction of detailed shape parameters and the construction of comprehensive size and morphology distributions for complex particulate systems.

We derive a new model of circadian oscillations in Neurospora crassa, which is suitable to analyze both temporal and spatial dynamics of proteins responsible for mechanism of rythms. The model is based on the non-linear interplay between proteins FRQ and WCC which are products of transcription of frequency and white collar genes forming a feedback loop comprised both positive and negative elements. The main component of oscillations mechanism is supposed to be time-delay in biochemical reactions of transcription. We show that the model accounts for various features observed in Neurospora’s experiments such as entrainment by light cycles, phase shift under light pulse, robustness to action of fluctuations and so on. Wave patterns excited during spatial development of the system are studied. It is shown that the wave of synchronization of biorythms arises under basal transcription factors.

This paper addresses the problem of detecting small objects in visible-spectrum aerial imagery. High background variability and weak feature saliency make small object detection a non-trivial task. Under such conditions, classical computer vision algorithms based on hand-crafted descriptors exhibit low efficiency, prompting a shift towards deep neural network architectures, which demonstrate superior generalization capability and robustness to false positives. We selected a one-stage approach based on a neural network predictive model as the primary object detection method. Also, several neural network architectures belonging to this class were analyzed, outlining their advantages and disadvantages. As the baseline detector, we adopted YOLO version 11 and incorporated a multi-scale feature aggregation module (which combines information from neural network layers operating at different scales) and a dimension-aware selective integration module (which automatically determines the dimension (channel, height, or width) along which to process features and fuses them selectively). These modifications aim to both enhance computational efficiency, enabling deployment of neural network models onboard aerial vehicles for real-time image analysis, and improve small object detection accuracy. Given the complexity of image annotation, we used an open synthetic image database containing approximately 4 000 images for training and testing (with a 90/10 split, respectively). We extend the training set using various random augmentation techniques. To evaluate the performance of the resulting predictive models, we employed mean Average Precision across all classes, using both a fixed 50% intersectionover- union threshold and a varying threshold from 50% to 95%. Overfitting was monitored by analyzing loss curves during training process. The proposed modifications to the YOLO architecture reduced image processing time by a factor of two while maintaining detection accuracy.

This paper formulates and solves a practical problem of data recovery regarding the distribution of languages on regional level in context of China. The necessity of this recovery is related to the problem of the determination of the linguistic diversity indices, which, in turn, are used to analyze empirically and to predict sources of social and economic development as well as to indicate potential conflicts at regional level. We use Ethnologue database and China census as the initial data sources. For every language spoken in China, the data contains (a) an estimate of China residents who claim this language to be their mother tongue, and (b) indicators of the presence of such residents in China provinces. For each pair language/province, we aim to estimate the number of the province inhabitants that claim the language to be their mother tongue. This base problem is reduced to solving an undetermined system of algebraic equations. Given additional restriction that Ethnologue database introduces data collected at different time moments because of gaps in Ethnologue language surveys and accompanying data collection expenses, we relate those data to a single time moment, that turns the initial task to an ’ill-posed’ system of algebraic equations with imprecisely determined right hand side. Therefore, we are looking for an approximate solution characterized by a minimal discrepancy of the system. Since some languages are much less distributed than the others, we minimize the weighted discrepancy, introducing weights that are inverse to the right hand side elements of the equations. This definition of discrepancy allows to recover the required variables. More than 92% of the recovered variables are robust to probabilistic modelling procedure for potential errors in initial data.

Analysis and practical aspects of implementing developed in the control theory robust control methods in studying economic systems is carried out. The main emphasis is placed on studying results obtained for dynamical systems with structured uncertainty. Practical aspects of implementing such results in control of economic systems on the basis of dynamical models with uncertain parameters and perturbations (stabilization of price on the oil market and inflation in macroeconomic systems) are discussed. With the help of specially constructed aggregate model of oil price dynamics studied the problem of finding control which provides minimal deviation of price from desired levels over middle range period. The second real problem considered in the article consists in determination of stabilizing control providing minimal deviation of inflation from desired levels (on the basis of constructed aggregate macroeconomic model of the USA over middle range period).

Upper levels of parameters uncertainty and control laws guaranteeing stabilizability of the real considered economic systems have been found using the robust method of control with structured uncertainty. At the same time we have come to the conclusion that received estimates of parameters uncertainty upper levels are conservative. Monte-Carlo experiments carried out for the article made it possible to analyze dynamics of oil price and inflation under received limit levels of models parameters uncertainty and under implementing found robust control laws for the worst and the best scenarios. Results of these experiments show that received robust control laws may be successfully used under less stringent uncertainty constraints than it is guaranteed by sufficient conditions of stabilization.

The aim of the paper is to obtain a numerical solution for linear and higher-order delay differential equations (DDEs) using the coded differential transform method (CDTM). The CDTM is developed and applied to delay problems to show the efficiency of the proposed method. The coded differential transform method is a combination of the differential transform method and Mathematica software. We construct recursive relations for a few delay problems, which results in simultaneous equations, and solve them to obtain various series solution terms using the coded differential transform method. The numerical solution obtained by CDTM is compared with an exact solution. Numerical results and error analysis are presented for delay differential equations to show that the proposed method is suitable for solving delay differential equations. It is established that the delay differential equations under discussion are solvable in a specific domain. The error between the CDTM solution and the exact solution becomes very small if more terms are included in the series solution. The coded differential transform method reduces complex calculations, avoids discretization, linearization, and saves calculation time. In addition, it is easy to implement and robust. Error analysis shows that CDTM is consistent and converges fast. We obtain more accurate results using the coded differential transform method as compared to other methods.

Eco-epidemiological models provide insights into factors influencing disease transmission and host population stability. This study developed two eco-epidemiological models to investigate the impacts of prey refuge availability and an Allee effect on dynamics. Model A incorporated these mechanisms, while model B did not. Both models featured predator – prey and disease transmission and were analyzed mathematically and via simulation. Model equilibrium states were examined locally and globally under differing parameter combinations representative of environmental scenarios. Model A and B demonstrated globally stable conditions within certain parameter ranges, signalling refuge and Allee effect terms promote robustness. Moreover, model A showed a higher potential toward extinction of the species as a result of incorporating the Allee effect. Bifurcation analyses revealed qualitative shifts in behavior triggered by modifications like altered predation mortality. Model A manifested a transcritical bifurcation indicating critical population thresholds. Additional bifurcation types were noticed when refuge and Allee stabilizing impacts were absent in model B. Findings showed disease crowding effect and that host persistence is positively associated with refuge habitat, reducing predator – prey encounters. The Allee effect also calibrated stability via heightened sensitivity to small groups. Simulations aligned with mathematical predictions. Model A underwent bifurcations at critical predator death rates impacting prey outcomes. This work provides a valuable framework to minimize transmission given resource availability or demographic alterations, generating testable hypotheses.

Dam safety assessment increasingly relies on continuous monitoring of hydrometeorological variables; however, identifying early-stage instability remains challenging due to complex spatio-temporal interactions and highly imbalanced failure observations. This study proposes a deep learning framework based on a Convolutional Bidirectional Gated Recurrent Unit (CNN–BiGRU) architecture to learn spatio-temporal precursors of dam instability from multivariate hydrometeorological time series. The convolutional component extracts localized temporal patterns associated with short-term fluctuations, while the bidirectional recurrent structure captures long-range dependencies and evolving dynamics preceding critical states.

The proposed model is evaluated on a real-world dam monitoring dataset comprising multiple water-level, meteorological, and derived dynamic indicators. To address class imbalance, a cost-sensitive training strategy using class weighting is adopted without synthetic oversampling. Experimental results demonstrate strong predictive performance, achieving an accuracy of 0.961, precision of 0.901, recall of 0.757, and an F1-score of 0.823. The model further attains a ROC-AUC of 0.907 and a PR-AUC of 0.819, indicating robust discrimination capability under imbalanced conditions.

Feature importance analysis reveals that short- and medium-term water level variability, including rolling standard deviation, volatility, and multi-scale gradients, play a dominant role in characterizing pre-instability behavior, providing physically interpretable insights into dam response dynamics. The findings suggest that the CNN–BiGRU framework effectively captures meaningful spatio-temporal precursors and offers a reliable data-driven tool for supporting dam safety monitoring and decision-making under real operational conditions.