This paper presents Multi_mil, a multilingual annotated corpus designed for the analysis of information operations in military discourse. The corpus consists of 1000 texts collected from social media and news platforms in Russian, Kazakh, and English, covering military and geopolitical narratives. A multi-level annotation scheme was developed, combining entity categories (e.g., military terms, geographical references, sources) with pragmatic features such as information operation type, emotional tone, author intent, and fake claim indicators. Annotation was performed manually in Label Studio with high inter-annotator agreement ($\mathsf{\kappa }$ = 0.82). To demonstrate practical applicability, baseline models and the proposed Onto-IO-BERT architecture were tested, achieving superior performance (macro-F1 = 0.81). The corpus enables the identification of manipulation strategies, rhetorical patterns, and cognitive influence in multilingual contexts. Multi_mil contributes to advancing NLP methods for detecting disinformation, propaganda, and psychological operations.

Ischemic stroke remains one of the leading causes of mortality and disability, and accurate segmentation of the affected areas on CT brain images plays a crucial role in timely diagnosis and clinical decision-making. This study proposes an ensemble approach based on the combination of the transformer-based models SE-UNETR and Swin UNETR using a weighted voting strategy. Its performance was evaluated using the Dice similarity coefficient, which quantifies the overlap between the predicted lesion regions and the ground-truth annotations. In this study, three-dimensional CT scans of the brain from 98 patients with a confirmed diagnosis of acute ischemic stroke were used. The data were provided by the International Tomography Center, SB RAS. The experimental results demonstrated that the ensemble based on transformer models significantly outperforms each individual model, providing more stable and accurate predictions. The final Dice coefficient reached 0.7983, indicating the high effectiveness of the proposed approach for ischemic lesion segmentation in CT images. The analysis showed more precise delineation of ischemic lesion boundaries and a reduction in segmentation errors. The proposed method can serve as an effective tool in automated stroke diagnosis systems and other applications requiring high-accuracy medical image analysis.

In this paper, the problem of global practical output tracking is investigated by state feedback for a class of uncertain nonlinear time-delay systems. Under mild conditions on the system nonlinearities involving time delay, we construct a homogeneous state feedback controller with an adjustable scaling gain. By a homogeneous Lyapunov-Krasovskii functional, the scaling gain is adjusted to dominate the time-delay nonlinearities bounded by homogeneous growth conditions and render the tracking error can be made arbitrarily small while all the states of the closed-loop system remain to be bounded.

Domain ontologies facilitate the organization, sharing, and reuse of subject areas. Building a software ontology is labor-intensive and time-consuming. In the process of obtaining a software ontology, it is required to analyze a huge number of scientific publications relevant to the software being modeled. The process of ontology replenishing with information from a huge number of scientific publications can be facilitated and accelerated through the use of lexical-syntactic patterns of ontological design. In this paper, we consider the possibility of automated construction of scientific subject area ontologies based on a heterogeneous patterns system of ontological design. This system includes ontological design patterns and is intended for ontology developers. System also includes automatically built lexical and syntactic patterns, which help to automatic replenishment of the ontology with information extracted from natural language texts.

Computer Simulation of Control of High-Order Nonlinear Systems using Feedback

Face recognition is a key area in computer vision and artificial intelligence. With the advent of deep learning, novel methods have been developed that achieve high accuracy in this field. This study aims to enhance facial recognition accuracy by comparing two state-of-the-art algorithms – DeepFace and FaceNet – and proposing an ensemble approach that integrates their strengths. To this end, we conduct a comparative analysis of DeepFace’s deep convolutional neural network–based identification and FaceNet’s multidimensional vector embedding, then combine their output probabilities into a single ensemble model. Experimental evaluation on publicly available datasets under varying lighting conditions and head poses reveals that the ensemble consistently outperforms each individual algorithm, demonstrating superior accuracy and robustness to external factors. We conclude that this hybrid ensembling strategy significantly improves recognition performance, validating its potential for complex face matching tasks. These findings indicate that real-world applications – such as surveillance, identity verification, and biometric authentication – can benefit from adopting ensemble methods to achieve higher accuracy and resilience.