Accurate transaction-level product detection under strict hardware and sensor limitations is necessary for top-loading smart vending cabinets. In many real-world implementations, the cabinet lacks weight sensors, RFID tags, and shelf-level instrumentation in favor of just two asynchronous cameras. Because they are inherently unable to describe intermediate product mobility, conventional snapshot-difference techniques omit pickup, return, and inspection events. Instead, they compare a small number of frames at the start and finish of a transaction. This paper proposes ZAB-Fusion, an edge-oriented multi-camera tracking framework for sensorless smart vending cabinets. The framework projects product tracks into a three-zone vertical cabinet model, integrates per-camera event streams using a strict cross-camera consensus rule, interprets compressed zone sequences using a finite-state event classifier, and combines a YOLO11-seg and RT-DETR detection ensemble with ByteTrack temporal association. Eight in-service vending transactions with eighteen ground-truth product pickups from seven different product classes are used to test the suggested approach. The released ZAB-Fusion configuration maintains a precision of 0.750 while improving recall from 0.389 to 0.500 and F1-score from 0.560 to 0.600 when compared to the snapshot-difference baseline. Zone categorization, finite-state interpretation, and cross-camera fusion are examples of symbolic reasoning components that contribute less than 3% of the total processing time, according to runtime study on an Intel N100 CPU-only edge device. These findings show that sensorless retail transaction recognition can be enhanced without retraining the underlying detector by using explicit motion semantics and auditable cross-camera consensus. ​

The high-level integration of generative artificial intelligence (AI) in edge computing systems has raised the question of the integrity and reliability of deploying Model-as-a-Service. Edge servers are not required to follow the so-called generative model to minimize computational cost, whereas users and service providers want validation mechanisms that do not compromise proprietary model information. To address this challenge, this study proposes a cooperative unmanned aerial vehicle (UAV)-swarm-enabled zero-knowledge verification framework for secure, privacy-preserving verification of edge-based generative artificial intelligence inference. The proposed framework involves edge servers producing an interactive cryptographic zero-knowledge proof to verify the execution of generative AI, and UAV swarms that fly freely to confirm verification operations, subject to mobility and energy constraints. The age of verification metric is proposed to trust verification information, jointly reflecting the unverified server reliability and verification freshness, and to provide dynamic priority to risky edge servers. To effectively plan the behaviour of a UAV swarm, a trust-based multi-agent reinforcement learning approach is developed that enables decentralized decision-making while training is centralized. Extensive simulation results show that the proposed framework significantly improves the state-of-the-art baseline schemes in verification timeliness, malicious server detection delay, energy efficiency, and scalability. The findings validate that integrating cooperative UAV swarms, trust-aware verification, and multi-agent learning is an efficient approach to providing reliable generative AI services in dynamic edge computing environments.

A novel path-planning method utilizing the trapezoid, adjacent, and distance, (TAD) characteristics of frontiers is presented in this work. The method uses the mobile robot’s sensor range to detect frontiers throughout each exploration cycle, modifying them at regular intervals to produce their parameters. This well-thought-out approach makes it possible to choose objective points carefully, guaranteeing seamless navigation. The effectiveness and applicability of the suggested approach with respect to exploration time and distance are demonstrated by empirical validation. Results from experiments show notable gains over earlier algorithms: time consumption decreases by 10% to 89% and overall path distance for full investigation decreases by 12% to 74%. These remarkable results demonstrate the efficacy of the suggested approach and represent a paradigm change in improving mobile robot exploration in uncharted territory. This research introduces a refined algorithm and paves the way for greater efficiency in autonomous robotic exploration. This study opens the door for more effective autonomous robotic exploration by introducing an improved algorithm.

Background/Objectives: Breast cancer diagnosis using histopathological images remains a critical yet challenging task in computational pathology due to overlapping visual features between benign and malignant tissues, inconsistencies in staining, and variations in magnification. The objective of this study was to design a lightweight yet high-performing deep learning model that bridges the gap between diagnostic accuracy and computational efficiency. Methods: We propose CellSage, a novel convolutional neural network (CNN) architecture enhanced with attention mechanisms. It integrates three core modules: a multi-scale feature extraction unit designed to capture both global and local tissue patterns; a depthwise separable convolution block for reducing computational load; and a Convolutional Block Attention Module (CBAM) to dynamically focus on diagnostically relevant regions. The model was trained and evaluated on the BreakHis dataset, using stain normalization (via contrastive augmentation modeling, CAM) and extensive data augmentation techniques. A patient-wise cross-validation strategy was employed to ensure robust generalization. Results: CellSage achieved 94.8% accuracy, a 0.93 F1 score, and a 0.96 AUC, while remaining compact at only 3.8 million parameters. It outperformed deeper and larger models such as ResNet-50, DenseNet-121, and Vision Transformers in terms of both predictive performance and computational efficiency. Ablation studies confirmed that multi-scale feature extraction and attention refinement were critical components. Conclusions: CellSage is an interpretable, reliable, and computationally lightweight system for breast cancer diagnosis using histopathological data. Its efficiency and low computational footprint render it an ideal candidate for real-time deployment on digital pathology platforms, particularly in environments with limited computational infrastructure.

In recent years, agricultural landscapes have increasingly suffered from severe fire incidents, posing significant threats to crop production, economic stability, and environmental sustainability. Timely and precise detection of fires, especially at their incipient stages, remains crucial to mitigate damage and prevent ecological degradation. However, conventional detection methods frequently fall short in accurately identifying small-scale fire outbreaks due to limitations in sensitivity and response speed. Addressing these challenges, this research proposes an advanced fire detection model based on a modified Detection Transformer (DETR) architecture. The proposed framework incorporates an optimized ConvNeXt backbone combined with a novel Feature Enhancement Block (FEB), specifically designed to refine spatial and contextual feature representation for improved detection performance. Extensive evaluations conducted on a carefully curated agricultural fire dataset demonstrate the effectiveness of the proposed model, achieving precision, recall, mean Average Precision (mAP), and F1-score of 89.67%, 86.74%, 85.13%, and 92.43%, respectively, thereby surpassing existing state-of-the-art detection frameworks. These results validate the proposed architecture's capability for reliable, real-time identification, offering substantial potential for enhancing agricultural resilience and sustainability through improved preventive strategies.

Rapid urban expansion has heightened the demand for accurate, scalable, and real-time methods to assess tree health and the provision of ecosystem services. Urban trees are the major contributors to air-quality improvement and climate change mitigation; however, their monitoring is mostly constrained to inherently subjective and inefficient manual inspections. In order to break this barrier, we put forward a lightweight multimodal deep-learning framework that fuses RGB imagery with environmental and biometric sensor data for a combined evaluation of tree-health condition as well as the estimation of the daily oxygen production and CO2 absorption. The proposed architecture features an EfficientNet-B0 vision encoder upgraded with Mobile Inverted Bottleneck Convolutions (MBConv) and a squeeze-and-excitation attention mechanism, along with a small multilayer perceptron for sensor processing. A common multimodal representation facilitates a three-task learning set-up, thus allowing simultaneous classification and regression within a single model. Our experiments with a carefully curated dataset of segmented tree images accompanied by synchronized sensor measurements show that our method attains a health-classification accuracy of 92.03% while also lowering the regression error for O2 (MAE = 1.28) and CO2 (MAE = 1.70) in comparison with unimodal and multimodal baselines. The proposed architecture, with its 5.4 million parameters and an inference latency of 38 ms, can be readily deployed on edge devices and real-time monitoring platforms.

Several crucial system design and deployment decisions, including workload management, sizing, capacity planning, and dynamic rule generation in dynamic systems such as computers, depend on predictive analysis of resource consumption. An analysis of the computer components' utilizations and their workloads is the best way to assess the performance of the computer's state. Especially, analyzing the particular or whole influence of components on another component gives more reliable information about the state of computer systems. There are many evaluation techniques proposed by researchers. The bulk of them have complicated metrics and parameters such as utilization, time, throughput, latency, delay, speed, frequency, and the percentage which are difficult to understand and use in the assessing process. According to these, we proposed a simplified evaluation method using components' utilization in percentage scale and its linguistic values. The use of the adaptive neuro-fuzzy inference system (ANFIS) model and fuzzy set theory offers fantastic prospects to realize use impact analyses. The purpose of the study is to examine the usage impact of memory, cache, storage, and bus on CPU performance using the Sugeno type and Mamdani type ANFIS models to determine the state of the computer system. The suggested method is founded on keeping an eye on how computer parts behave. The developed method can be applied for all kinds of computing system, such as personal computers, mainframes, and supercomputers by considering that the inference engine of the proposed ANFIS model requires only its own behavior data of computers' components and the number of inputs can be enriched according to the type of computer, for instance, in cloud computers' case the added number of clients and network quality can be used as the input parameters. The models present linguistic and quantity results which are convenient to understand performance issues regarding specific bottlenecks and determining the relationship of components.

Exploration of mobile robot without prior data about environments is a fundamental problem during the SLAM processes. In this work, we propose improved version of previous Rmap algorithm by modifying its Exploration submodule. Despite the previous Rmap's performance which significantly reduces the overhead of the grid map, its exploration module costs a lot because of its rectangle following algorithm. To prevent that, we propose a new Rmap+ algorithm for autonomous path planning of mobile robot to explore an unknown environment. The algorithm bases on paired frontiers. To navigate and extend an exploration area of mobile robot, the Rmap+ utilizes the inner and outer frontiers. In each exploration round, the mobile robot using the sensor range determines the frontiers. Then robot periodically changes the range of sensor and generates inner pairs of frontiers. After calculating the length of each frontiers' and its corresponding pairs, the Rmap+ selects the goal point to navigate the robot. The experimental results represent efficiency and applicability on exploration time and distance, i.e., to complete the whole exploration, the path distance decreased from 15% to 69%, as well as the robot decreased the time consumption from 12% to 86% than previous algorithms.

The article presents the results of studies on the study of the contact zone of masonry mixtures, with the aim of developing compositions for the construction of walls and partitions made of bricks and blocks in seismically active regions. Analyzes of the developed compositions, theoretical and practical substantiation of high adhesive capacity in conditions of high seismic activity are presented. Methods of using non-fired binders based on high-modulus silicate components are proposed. The technological scheme and methods of obtaining universal masonry mortars providing high adhesive capacity and adhesion strength are presented. When studying the “brick–mortar” contact zone by X-ray thermal electron probe microanalysis, a regular relationship between the chemical compounds of the system was established and the correlation dependence of the micro hardness of neoplasm’s in the contact zone was determined. It has been established that an increased concentration of sodium, aluminum, silicon is observed in the contact zone, which contribute to the intensification of the formation of a dense microstructure due to the presence in it of water-resistant neoplasm’s such as low-basic hydro silicates and alkaline hydro aluminates. These new formations constitute the structural matrix (frame) of the contact zone and play the role of a damper at the time of the action of alternating seismic loads. The developed compositions have found application as universal masonry mortars in the construction of brick walls with the provision of category I adhesion strength.

Navigation in the absence of initial environmental information is a situation in which a robot is faced with the difficulty of traversing an unknown area for exploration with obtaining the environmental information simultaneously. Therefore, to complete and optimize the exploration efficiently, the robot needs an autonomous path-planning algorithm. This work proposes a new autonomous path-planning algorithm for exploration in an unknown environment based on paired frontiers, which we call internal and external frontiers algorithm (IEFA), that defines extended area for navigation of the mobile robot. For each exploration round, the robot defines external frontiers using the maximum range of sensors. Then, the robot generates internal frontiers, that is, pairs of external frontiers by varying the range of sensors. According to the size of each pair of frontiers, the algorithm generates the target point for robot navigation. The frontiers of internal layer are utilized as a main parameter for generation of next exploration point. We evaluated the proposed algorithm in simulation environments using the ROS toolbox of MATLAB and compared it with two previous exploration algorithms. From the experimental results, the proposed algorithm showed from 31% to 85% better performance in the path distance than previous algorithms.

The article presents data on the results of research and processing of phosphogypsum waste from the Samarkand Chemical Plant, over many years stored on a vast territory. Phosphogypsum contains up to 94% contains two water sulfates of calcium, which is the most valuable raw material to produce gypsum binders. However, phosphogypsum contains some chemical compounds that are harmful to human life and for this reason, it is difficult to use in industrial production without neutralization. The article presents the results of research and recommendations for solving this problem.

The monitoring utilization and workloads of computer hardware components, such as CPU, RAM, bus, and storage, are an ideal way to evaluate the effectiveness of these components. In this paper, we surveyed the basic concepts, characteristics, and parameters of computer systems that determine system performance, and the types of models that provide adequate modeling of these systems. We investigated and developed the applied aspects of the theory of fuzzy sets’ principles and the Matlab environment tools for monitoring and evaluating the state of computing systems. The idea of the paper is to identify the state of the computer infrastructure by using the models of Mamdani and Sugeno FIS (fuzzy inference system) to evaluate the impact of RAM and storage on CPU performance. With this approach, we observed the behavior of computer infrastructure. The results are useful for understanding performance issues with regard to specific bottlenecks and determining the correlation of performance counters. Moreover, the model presents linguistic results. Hereafter, performance counter correlations will support the development of algorithms that can detect whether the performance of a given computer will be affected by a reasonable priority. The performance assertions derived from these approaches allow resource management policies to prevent performance degradation, and as a result, the infrastructure will be able to serve safely as expected. These methods can be applied across the entire spectrum of computer systems, from personal computers to large mainframes and supercomputers, including both centralized and distributed systems. We look forward to their continued use, as well as their improvement when it is necessary to evaluate future systems.

In this paper, we suggest an activity and health monitoring system to observe the status of the dogs in real time. We also propose a k-days algorithm which helps monitoring pet health status using classified activity data from a machine learning approach. One of the best machine learning algorithm is used for the classification activity of dogs. Dog health status is acquired by comparing current activity calculation with passed k-days activities average. It is considered as a good, warning and bad health status for differences between current and k-days summarized moving average (SMA) > 30, SMA between 30 and 50, and SMA < 50, respectively.

In this paper we considered the steps of the method of accelerated digital signal processing algorithms with modern software tools parallel processing. Compared the experimental data, obtained during the processing of parallelizing OpenMP technology and Cilk Plus. As a hardware implementation architecture are analyzed multi-core processors.