This article addresses the problem of identifying unique objects in aerial images of urban areas on the Earth’s surface, which can serve as stable landmarks for UAV navigation without GPS signals. The main contribution lies in proposing an approach to transforming the image into an object-oriented vector representation (embedding) that retains structural information about those objects. The proposed approach automatically identifies the most distinctive objects, which can serve as navigation landmarks. The study focuses on urban and suburban landscapes, where buildings are chosen as landmarks and YOLOv11 is used as the deep learning model. By employing dimensionality reduction methods, in particular PCA and t-SNE, it is demonstrated that in the proposed embedding space, buildings with atypical structural or visual characteristics differ significantly from other buildings and are easily classified as outliers, making them natural landmarks for navigation. Experimental results confirm the effectiveness and potential of the proposed approach for ensuring stable UAV navigation in scenarios where GPS may be inaccessible—the accuracy of identifying buildings designated as landmarks is twice that of ordinary buildings (Recall@1 = 0.51 vs. 0.28).
Intelligent systems for optimized object placement in medical and biological applications leverage artificial intelligence advanced data fusion techniques to enhance precision, efficiency, and patient outcomes. These systems tackle a range of issues, including the positioning of surgical tools, deployment of sensors, and analysis of diagnostic images. Advanced mathematical modeling has become essential in healthcare and biological research, driving innovative solutions for treatment planning and spatial arrangements. This paper introduces an intelligent system aimed at optimizing the placement of geometric objects in medical and biological contexts. We employ a universal mathematical model that functions as an intelligent agent, utilizing parameters to adapt to different scenarios and optimize outcomes. We develop mathematical models and advanced algorithms to ensure precise placement, achieving the desired therapeutic or research outcomes while minimizing adverse effects. The mathematical model is formulated as a knapsack problem and expressed as Mixed Binary Non-Linear Programming (MBNLP). Problems related to optimized object placement can be addressed by selecting different model parameters. Several implementations demonstrate this approach, including Gamma Knife radiosurgery, laser coagulation, brachytherapy, and chromosome territory modeling. These systems tackle a range of issues, including the positioning of surgical tools, deployment of sensors, and analysis of diagnostic images. Advanced mathematical modeling has become essential in healthcare and biological research, driving innovative solutions for treatment planning and spatial arrangements. This paper introduces a smart system aimed at optimizing the placement of geometric objects in medical and biological contexts.
This paper focuses on the development of an intelligent driver monitoring system based on adaptive deep learning models to enhance road safety. The research explores advanced deep learning techniques, particularly convolutional neural networks and their modifications, such as ResNet50 and MobileNetV2. Special attention is given to the stages of data preprocessing, augmentation, training and testing dataset formation, as well as model training and fine-tuning. A conceptual framework and architecture for an intelligent driver monitoring system have been developed, incorporating two modules based on different deep learning models. An experimental study was conducted to compare the performance of various convolutional neural network (CNN) architectures, including classical CNN, ResNet50, MobileNetV2, EfficientNetB0, and VGG16, in detecting driver fatigue and drowsiness. Signs of overfitting were identified in the ResNet50 and MobileNetV2 models when applied to the selected datasets, highlighting the need for further hyperparameter optimization. The developed testing scripts enable real-time analysis of behavioral indicators of drowsiness and driver distraction. The proposed system is designed for non-invasive and high-precision real-time monitoring of driver conditions, including fatigue, drowsiness, and distraction detection. The findings confirm the effectiveness of adaptive deep learning models for driver state monitoring. The developed system demonstrates the capability to detect signs of fatigue, drowsiness, and distraction, which may help reduce the likelihood of road accidents. Experimental results indicate that the choice of an optimal neural network architecture depends on the specific task requirements and the available computational resources.
In the paper, a proposed approach for improving efficiency of web propaganda patterns detection by transformer neural networks is presented. Approach consists of sequential use of three developed methods: method for dataset balancing, method for fine-tuning individual binary neural network models and method for detecting web propaganda patterns. Compared to existing analogues, the use of proposed approach allowed achieving an efficiency increase of 0.1 by F1 metric when detecting propaganda patterns in web texts using transformer neural networks due to dataset balancing optimization. Analyzing the impact of parameter that determines proportion of texts without web propaganda patterns allows assessing how the models ability to distinguish propaganda patterns from neutral texts and texts with other propaganda patterns. This allows finding the optimal ratio of dataset classes to increase the overall effectiveness for detecting web propaganda patterns. Conducted research has established that the highest results were achieved when forming the training dataset with a percentage of texts without patterns of 30% using the RoBERTa neural network, and was achieved 0.725 by F1 metric. Proposed approach ensures the determination of the optimal ratio between text sets with propaganda patterns and neutral text set, which improving the generalization ability of models and reduce their bias.
This article presents an approach to automated penetration testing using a distributed multi-agent system (MAS) based on the JADE platform. The proposed architecture includes four specialized agents that collaborate to dynamically bypass Web Application Firewalls (WAFs) and Intrusion Prevention Systems (IPS). The system enables the real-time generation of polymorphic payloads, reinforcement learning-based adaptation of attack strategies, and distributed deployment to reduce detection risks. Experimental validation confirms that this multi-agent approach enhances the efficiency of vulnerability detection through parallel execution, intelligent payload mutation, and reduced reliance on manual testing.
This study investigates the application of artificial intelligence techniques for object segmentation in high- resolution satellite imagery, with a focus on the automatic classification of land cover types such as rivers, forests, and buildings. It includes a comparative analysis of traditional image processing methods and modern deep learning architectures — specifically convolutional neural networks (U-Net, DeepLabV3+, Mask R-CNN) and transformer-based models. The study outlines practical considerations for model deployment and highlights future directions, including the use of self-supervised learning, lightweight models for edge devices, and multi-modal data integration. The findings highlight the advantages of AI- driven segmentation over traditional methods, improving precision and scalability for applications in environmental monitoring, urban planning, and disaster management.
Diabetes Mellitus is a metabolic complex and chronic non-communicable disorder affecting a large population in the world. Different studies have shown the damage caused by Diabetes Mellitus on multiple systems, which leads to complications such as cancer, cardiovascular disorders, and sarcopenia. The changes in insulin, glycaemia, or glucose levels bring multiple changes in the body, including the formation of oxidative species, inflammation, Advanced Glycation End (AGE) products, and hormonal imbalance. In recent times, more attention has been given to the association of Diabetes and cognitive dysfunction because of its increasing prevalence and the severe impact on the lives of diabetic patients. Moreover, the part of different proteins and pathways related to Diabetes that lead to the occurrence of other diseases has been demonstrated. This research presents a predictive model for the early detection of diabetes-associated cognitive diseases using machine learning techniques. The model utilizes patient health records, lifestyle factors, and diabetes progression data to predict cognitive decline risks. The dataset is pre-processed using statistical analysis, followed by feature selection techniques to optimize the model's performance. Various machine learning algorithms, including decision trees, random forests, and neural networks, are explored to determine the most accurate approach for predictive analysis. The study demonstrates that early detection models can effectively predict diabetes-associated cognitive decline (DACD) onset with high precision, offering a valuable tool for healthcare providers. The results show that predictive models can support timely interventions and personalized treatment plans for at-risk patients.
Rapid detection of forest fires is crucial to reduce their devastating impact on ecosystems and human lives. In this paper, we present an AI-based solution for forest fire detection using deep learning from satellite imagery using the ResNet50V2 convolutional neural network (CNN). The dataset used to train the model consists of 1,900 images (950 per class), carefully curated to reflect real-world scenarios of both active forest fires and undisturbed forests. Data preprocessing included image augmentation to reduce overfitting and enhance model performance. Transfer learning, model regularization, and reconstructed pooling layers were applied during training on this dataset, which was augmented with techniques such as random horizontal rotations, zooming, and cropping to improve model generalization. The model achieved 97.63% accuracy and 98.40% precision in detection. Forest fire detection using satellite images is very useful because CNN methods can detect and locate active fires more than once per hour. It is well known that the earlier a forest fire is detected, the more effective it is for people and the environment. This method can help to develop of new strategies for real-time fire monitoring systems, in addition to greatly enhancing wildfire management and prevention efforts. This study focuses not on early fire detection, but on identifying post-wildfire damage using deep learning techniques applied to satellite imagery.
This research presents the development of an intelligent controller for the helicopter turboshaft engines gas temperature, aimed at compensating for the measuring sensor’s inertial delays and optimizing transient processes. The aim is to compensate for inertial delays τ ≈ 0.025 seconds and optimize transient processes. The method is based on a double summation circuit with a channel selector, comparing signals from a thermocouple and a gas-generator rotor speed sensor, and an adaptive observer based on Pade approximation and Taylor series expansion provides a prediction of the state at t + τ. The intelligent control law includes a proportional-integral-differential structure with the coefficients γi correction via gradient descent. To refine the delay estimate, a two-layer fully connected multilayer perceptron with a SmoothReLU activation function is implemented trained on flight test data was implemented, which reduced τ to 0.016 seconds (–36 %). This module allows to approximate nonlinear relations between input features and the delay value, which ensures the control signal’s timely correction and the system’s adaptation to changing operating conditions. Modeling of the system in the Matlab Simulink environment demonstrated a significant improvement in the transient process characteristics: overshoot was reduced from 8.0 to 1.5 %, and the mode establishment time was reduced from 4.2 to 3.3 seconds. The neural network module testing showed high predicting accuracy (99.537 % with losses of 0.511 %), confirmed by the determination coefficient R2 = 0.9717. The neural network use made it possible to reduce the delay value to 0.016 seconds, which corresponds to an improvement of 36 % compared to traditional methods. The obtained results indicate a proposed technique’s high potential for improving the helicopter turboshaft engines automatic control system’s dynamic accuracy and stability.
Permutation entropy (PEn) is a widely adopted nonlinear statistical measure for quantifying complexity in time series data. Despite its conceptual clarity and computational efficiency, classical PEn has notable limitations, particularly its disregard for amplitude variations in time series data and the simplistic handling of sequences containing equal-valued observations. Although modified PEn methods exist, their potential as early-warning indicators for cryptocurrency market crashes remains largely unexplored. This paper addresses these limitations by conducting a comparative analysis of classical PEn and three enhanced methods: weighted permutation entropy (WPEn), amplitude-aware permutation entropy (AAPEn), and uniform quantization-based permutation entropy (UPEn). Specifically, these entropy metrics are employed to analyze the Bitcoin market crash from December 2017 to February 2018, utilizing a sliding window approach. Empirical results demonstrate that amplitude-enhanced entropy methods effectively capture nuanced market dynamics and fluctuations, offering more precise and more reliable signals of impending market instability. This study confirms the value of advanced entropy measures in cryptocurrency markets and underscores their potential as robust indicators for detecting and forecasting financial crashes.
The rapid advancements in generative adversarial networks (GANs) have significantly impacted digital content synthesis, presenting both opportunities and challenges in multimedia forensics and cybersecurity. We present an Enhanced Adaptive DCGAN (EADC-GAN) for generating high-fidelity synthetic fingerprints, addressing core challenges in training stability and sample diversity. By combining Wasserstein loss with gradient penalty (WGAN-GP), instance normalization in the discriminator, and tailored architectural refinements, our model achieves strong image realism at reduced training cost. Compared to prior DCGAN- based methods, EADC-GAN synthesizes more diverse, artifact-free samples in fewer epochs, making it suitable for scalable biometric data generation. This has key implications for secure authentication, privacy- preserving biometric datasets, and adversarial robustness in cybersecurity contexts.
Predictive control plays a significant role in mobile robotics, especially in trajectory tracking, obstacle avoidance, and real-time decision-making. In this study, we explore how Markov Decision Processes (MDPs) can be integrated with predictive control to enhance navigation, particularly in maze-like environments. A case study on MDP-based maze exploration analyzes key system limitations, including computational complexity and real-time adaptability. While MDPs often struggle to adapt to dynamic environments, predictive techniques like Model Predictive Control (MPC) offer improvements in trajectory optimization and responsiveness. We also discuss practical applications in areas such as warehouse navigation and multi- robot coordination, showing the benefits of combining MDPs and predictive control for robust performance in real-world scenarios.
Currently, the development of discrete signal filtering methods that are used in computerized biometric identification systems is an urgent task. In the case of linear filtering, one of the most popular methods is the Kalman filter. A modified Kalman filtering method was proposed to improve the efficiency of digital filtering of a discrete signal. This method provides automation of parameter value determination and improves the speed and accuracy of Kalman filtering by using fewer parameters and identifying them based on immune metaheuristic methods. The proposed metaheuristic methods reduce the probability of convergence to a local extremum by using the Cauchy distribution and make parametric identification more accurate. Algorithms of immune metaheuristic methods for identifying Kalman filter parameters have been developed, which are designed for software implementation on the GPU using CUDA technology, which increases the accuracy of Kalman filtering. Further prospects of the study are to utilize the proposed immune metaheuristic methods for various general and special purpose intelligent systems.
This study evaluates the effectiveness of physics-informed neural networks (PINNs) for solving both stationary and non-stationary partial differential equations (PDEs), including those with Robin boundary conditions, in rectangular and non-rectangular domains. Although only a small subset of the PINN literature examines mixed boundaries or tackles non-rectangular geometries, and even fewer studies benchmark accuracy against the finite-element method (FEM), the present work provides precisely that comparison. We test feedforward PINNs with tanh activation, whose depth and width were empirically selected for each benchmark to balance accuracy and training cost; training uses Adam optimization with Glorot initialization. These networks are evaluated on three problems: a 2-D Laplace equation on a square (Dirichlet–Neumann– Robin), the same equation on a doubly connected domain with Dirichlet boundary conditions, and a 1-D non- stationary heat equation with Robin boundaries. A weighted mean-squared residual, evaluated via automatic differentiation in TensorFlow, balances equation, boundary, and initial-time terms, thereby handling non- stationary problems without a separate time-stepping scheme. Within the tested class, linear, second-order parabolic and elliptic PDEs in 1-D and 2-D, the network attains ≤ 3 % l ∞ error relative to analytical or FEM solutions after 4–6 min of training on an RTX 3080 Ti Laptop GPU, matching FEM accuracy while eliminating meshing and easing equation and boundary changes. The time to compute a standard PINN solution is longer than for a FEM solution for problems considered in the research, and a broad literature review reveals theoretical convergence limits that constrain standard PINNs to modest-scale, well-conditioned diffusion problems.
Single Image Super-resolution (SISR) methods are actively developed with the help of advancements in Convolution Neural Networks (CNNs) and attention mechanisms. Following the progress in RGB SISR methods, thermal image super-resolution methods (TISR) are beginning to adopt and implement these advancements. Despite showing prominent results, modern state-of-the-art SISR methods often have a large number of parameters, leading to a significant computational overhead and memory consumption and making it difficult to run these methods in real-time or on edge devices. To address these problems, we propose a parameter-efficient TISR model named LECAN, which consists of a stack of efficient channel- spatial attention blocks (ECSAB). Specifically, the ECSAB combines Pixel Attention (PA) with the proposed Efficient Contrast-aware Channel Attention (ECCA) to extract both spatial and channel-wise features while maintaining a low parameter count. Meanwhile, the Attentive Feature Fusion (AFF) mechanism effectively combines information from all blocks, capturing both low-level and high-level features. The qualitative and quantitative results show that the proposed method achieves superior results among same-size models while preserving the texture and patterns of the thermal image with a small number of parameters.
This study focuses on improving the performance and scalability of web applications by leveraging modern architectural solutions and optimization techniques. The research explores key aspects such as the Model-View-Controller (MVC) pattern, process clustering, and database performance under high loads. The researchers conducted experimental tests to evaluate system efficiency and analyzed load-balancing algorithms to improve scalability. The study resulted in developing a web application that supports MVC, clustering, and Cross-Origin Resource Sharing (CORS), demonstrating its practical applicability in educational platforms, e-commerce, and CRM systems. The implemented solution performed efficiently under high-load conditions, significantly improving response times and handling multiple simultaneous requests. The findings emphasize the importance of modern optimization techniques in ensuring high performance, improving user experience, and increasing conversion rates.
This study investigates the use of EfficientNet-B0, a computationally efficient convolutional neural network architecture, for satellite image classification using the EuroSAT dataset. Evaluating the baseline EfficientNet-B0 model, we achieved 98.1% overall accuracy and a 0.98 macro-averaged F1 score on the test set. This performance is highly competitive with state-of-the-art results reported for the EuroSAT dataset, demonstrating that EfficientNet-B0 offers a strong balance between high accuracy and computational efficiency. These findings suggest that EfficientNet-B0 is a promising approach for tasks requiring efficient satellite image categorization, such as large-scale land use monitoring, urban planning, and environmental management analysis based on satellite imagery.
The research about implementation transfer learning in medical diagnostics is important because it allows to evaluate how well already trained neural networks can adapt to specific medical data. This helps to understand which architectures work best, how to improve diagnostic accuracy, and reduce the risk of false positives. In addition, such research contributes to the development of more reliable and interpretable models, which is critical for physician confidence and the implementation of AI in real-world clinical practice.
The focus of this research is the creation of intelligent geometric design technologies. The system employs state-of-the-art methods and tools to automate the arrangement and enhance the placement of 3D shapes. Specifically, the aim is to resolve practical issues in optimizing additive manufacturing processes. This is accomplished by merging artificial intelligence techniques with novel computational solutions for superior results. The article presents a nonlinear optimization approach for solving 3D irregular packing problems with arbitrarily moved and rotated objects. Phi-functions and quasi-Phi- functions are used to describe interactions between the 3D objects. The following formulation presents the packing problem in mathematical terms, along with an analysis of its features. A local optimization algorithm is introduced to identify solutions, with a focus on the characteristics that have been delineated. The results of computational experiments suggest that the proposed solution method is effective for 3D irregular packing optimization.
This study explores the concepts of computational optimization heuristics for item categorization and allocation on distribution center racks. Properly organizing products on racks in warehouses or distribution centers according to their respective categories is essential for efficient operations. The research proposes the flower-cutting optimization heuristics with 17 tuning parameters. Special attention is given to examining the effect of two grouping tuning parameters on the number of generated product allocations and their implications for space utilization, accessibility, and operational efficiency: implements a grouping strategy where, for each total width, only one product allocation with the maximum total profit is considered; adopts a grouping strategy where, for each total profit and profit ratio, only one product allocation with the minimum total width is selected. The experiment revealed that the implementation of grouping tuning parameters plays a crucial role in substantially reducing computational requirements while preserving the quality of solutions. By narrowing the solution space, these parameters ensure that the heuristics efficiently produce near-optimal allocations. This streamlined approach enhances the practicality of addressing large-scale shelf space allocation challenges, making the heuristics highly applicable to real-world scenarios.
