This paper presents a hybrid approach for accelerating 3D smoke simulations in computer graphics using a lightweight deep neural network for volumetric upscaling. High-fidelity smoke simulation is computationally intensive due to the cubic growth of spatial resolution and the cost of solving the Navier-Stokes equations. Our method reduces computational demand by performing simulation at a lower resolution and subsequently reconstructing high-resolution volumetric fields using a compact neural network architecture. The proposed model focuses on recovering fine-scale turbulent structures and visual complexity while preserving the large-scale physical consistency provided by the underlying solver. Experimental results indicate substantial computational speedups relative to native high-resolution simulation, while qualitative comparisons suggest plausible recovery of fine-scale smoke detail. The proposed framework therefore targets interactive or near-real-time smoke preview workflows on commodity hardware, improving accessibility for iterative graphics production.

High-resolution simulation of incompressible smoke is computationally expensive because it requires repeatedly enforcing incompressibility through large Poisson solves while advecting fields that contain fine-scale structure. Although classical Eulerian projection-based solvers are robust and widely used, their cost grows rapidly with grid resolution, making high-fidelity results difficult to achieve in interactive settings. Recent progress in deep learning suggests that expensive numerical operators can be approximated by compact neural surrogates. In this work, we accelerate the pressure projection step in a two-dimensional smoke solver by replacing the iterative Poisson solver with a lightweight neural model. Instead of learning the entire simulation pipeline, we target the dominant linear solve and learn a direct mapping from the divergence of the intermediate velocity field to the corresponding pressure field used for projection. Training data is generated directly from the baseline simulator using its existing discretization and boundary handling, ensuring consistency between supervision and deployment. The proposed network is a multi-scale transformer-based encoder-decoder designed for efficient inference. To preserve physical correctness, training uses a composite objective that combines pressure reconstruction with an operator-level constraint that penalizes violations of the Poisson equation. The learned solver is integrated as a drop-in module within the simulator and supports varying runtime resolutions through interpolation before and after inference. The resulting system provides substantial speed-ups of the projection stage while maintaining visually plausible behavior over the evaluated sequences, enabling faster high-resolution smoke generation without modifying the remaining components of the solver.

This study presents a data augmentation method based on angular-radial segmentation, which enables a balanced expansion of the point set while preserving the original spatial distribution. The procedure first involves determining the centroid of the set and shifting all observations. Next, the sample farthest from the origin is identified for the determination of a maximum base radius that will define the radius of the first segment computed in the subsequent step. The entire set is divided into circular segments with an angular width of α and radius R i which represents the distance of the farthest point within the i th segment from the origin. Within each segment subsections are defined in such way that each has an equal area within the i th segment. Once the areas have been determined, new points are distributed proportionally to the number of existing points in each subsection randomly in boundaries designated by each subsection. This ensures that the augmentation of new points is carried out in a manner consistent with the density of the respective subsections relative to the entire original set.

Decentralized tasks use machine learning models with certain assumptions. The first is the sharing of weights and feature extractors. The second is maintaining the privacy of the data. The idea of learning using multiple models can also be applied in parallel training, where a given model is trained on a different thread. This has applications in creating models based on federated learning, the Internet of Things and Digital Twins. This paper proposes a new neural network model that uses the atrous technique and attention modules. In addition, we propose a hierarchical-based training strategy, where the best model shares weights and is omitted during further training. This reduces the number of training epochs and increases the model's adaptability to a given set. The tests conducted on a publicly available medical database indicate high learning potential for both the proposed model and the hierarchical learning strategy.

Agricultural health monitoring is a critical task in ensuring the stability of modern agriculture. Many plant diseases share visual similarities, making manual inspection both time consuming and error prone, which is why robust and adaptable disease detection frameworks are not only desirable but essential to maintaining a resilient agricultural ecosystem. In this paper, we propose a hybrid federated learning (FL) framework that integrates a globally shared feature extractor with a client-specific self-attentive branch and classifier. The proposed framework uses a global model with both globally shared and client-tailored branches to achieve better performance for specialized tasks in decentralized training scenarios. The experiments were carried out on a Plant Village data set in a scenario, where each client represented a different crop type and faced a different leaf disease classification problem. The proposed solution revolved around the clients sharing the global weights, thus simultaneously contributing towards better feature extraction of the common leaf features, while the specialized segment of the model focused on proper interpretation of the extracted features (via cross-attention mechanism) and direct classification. The results obtained demonstrate the effectiveness of the proposed approach over standard local training, as training with the proposed hybrid FL framework resulted in a perfect classification of the precision 100% of apple leaf disease.

3D point clouds (LiDAR) are an integral application in robotics, creating 2D or 3D maps that help autonomous vehicles navigate or avoid and recognize obstacles as they drive. In this work, we propose Graph Convolutional Network with Multi-Scale Pooling to classify three-dimensional objects. Previous research very rarely undertakes the transformation of point clouds into graphs and the use of increasingly popular graph networks. Researchers focus on transforming 3D points into voxels and using classical CNN or 3D CNN convolutional networks or methods that directly use the entire cloud, as is evident in the PointNet model. Therefore, we focused on transforming our data into irregular structures (graphs) and classifying them. Our proposed method increases the possibilities for point cloud interpretation by taking into account the relationships between points in space. In order to achieve even better accuracy results, we use the integration of two pooling techniques in the final stage of the model. Our GCN with Multi-Scale Pooling architecture has been trained and tested on the ModeiNetl0 dataset, achieving high-quality metrics: Accuracy, Precision, Recall, and F1-Score compared to other classical SOTA models in the classification domain. Additionally, we carried out experiments based on ShapeNet Core dataset, for which our network achieved almost 100% accuracy. In addition, we showed that our proposed model is built from a much smaller number of parameters compared to other modern methods.

Deep learning has significantly improved image quality in enhancement and retouching tasks. However, current methods, such as HDRNet, CSRNet, and 3D LUT, primarily rely on low-level visual features and lack in-depth utilization of image semantic information, resulting in global average enhancement, color inconsistencies, and loss of brightness details. Noise and blur in images intensify the blending of distinct features and contribute to information degradation, making it more challenging to extract semantic details and thereby limiting the recovery performance. In order to solve the above problem, this paper proposes SUNet, an image restoration model enhanced with semantic information. By incorporating fine-grained semantic information into UNet and employing orthogonal and decoupled feature representations, SUNet significantly improves restoration performance without relying on specific segmentation annotations. The introduction of semantic information enables the model to differentiate between different regions in the image distinctly, making feature extraction more targeted and progressively reducing feature coupling. This enhances the model’s ability to represent semantically relevant features while avoiding interference from blurry or noisy features. Our contribution effectively bridges the gap between global enhancement techniques and the need for local semantic accuracy, laying the foundation for more sophisticated image enhancement methods. Experimental evaluations on public datasets demonstrate that our approach outperforms state-of-the-art methods.

Analysis of the immediate environment and its segmentation is particularly useful in the mechanism of autonomous vehicles. Accurate marking of important points and places while traversing the route allows cars to analyze the space and navigate safely. Transformers models and their successors such as UNETR and Vision Transformer using Multi-Head Attention are widespread in the segmentation process. These approaches permit to better image classification and segmentation results, as each head in the transformer simultaneously analyzes different aspects of the relationship between image fragments (patches). Given the high potential of transformer algorithms, in this article we present an extended UNETR model, which in its design additionally includes the RSAM attention block (Residual Spatial Attention Module) for the road segmentation process. Our RSAM-UNETR architecture supported by additional weights in the loss function performed perfectly and achieved Accuracy = 99.04%, mDice = 89.21% and mIoU = 81.80% with a very stable training process achieving a higher mDice value by 5.46% and mIoU by 7.31% compared to the classic UNETR.

Abstract The current proofreading algorithms for action regulation mainly recover the 3D structure and action information of non-rigid objects from image sequences by factorization. Most of algorithms assume that the camera model is an affine model. This assumption only holds if the size and depth of the object change very little relative to the distance from the object to the camera, which is in the case of fixed-shape basis. When the object is very close to the camera, this assumption causes a large reconstruction error. This paper solves this problem by the intelligent proofreading algorithms for remote skiing teaching actions based on variable shape basis. Firstly, the improved Retinex algorithm is used to enhance the multi-frame video images of skiing actions to make the action details more prominent. Then, measurement matrix is calculated after eliminating the translation vector by coordinate transformation. Under the condition of rank constraint, the measurement matrix is decomposed by singular value decomposition algorithm, and the correct shape basis structure of 3D action features can be obtained by using the variable shape basis. Finally, by randomly initializing a parameter, the optimized parameter and the least square algorithm are used to optimize the randomly initialized parameter further. The iteration until the convergence of the objective function can be used to calculate the deformation degree of the actions. The test results show that this algorithm improves the proofreading accuracy of action regulation in skiing teaching, and the proofreading results of various uploaded sliding actions are correct, which can be applied to remote skiing teaching and community learning.

Abstract Agriculture serves as the predominant driver of a country's economy, constituting the largest share of the nation's manpower. Most farmers are facing a problem in choosing the most appropriate crop that can yield better based on the environmental conditions and make profits for them. As a consequence of this, there will be a notable decline in their overall productivity. Precision agriculture has effectively resolved the issues encountered by farmers. Today's farmers may benefit from what's known as precision agriculture. This method takes into account local climate, soil type, and past crop yields to determine which varieties will provide the best results. The explainable artificial intelligence (XAI) technique is used with radial basis functions neural network and spider monkey optimization to classify suitable crops based on the underlying soil and environmental conditions. The XAI technology would provide assets in better transparency of the prediction model on deciding the most suitable crops for their farms, taking into account a variety of geographical and operational criteria. The proposed model is assessed using standard metrics like precision, recall, accuracy, and F1‐score. In contrast to other cutting‐edge approaches discussed in this study, the model has shown fair performance with approximately 12% better accuracy than the other models considered in the current study. Similarly, precision has improvised by 10%, recall by 11%, and F1‐score by 10%.

Machine learning has demonstrated tremendous potential in solving real-world problems. However, with the exponential growth of data amount and the increase of model complexity, the processing efficiency of machine learning declines rapidly. Meanwhile, the emergence of quantum computing has given rise to quantum machine learning, which relies on superposition and entanglement, exhibiting exponential optimisation compared to traditional machine learning. Therefore, in the paper, we survey the basic concepts, algorithms, applications and challenges of quantum machine learning. Concretely, we first review the basic concepts of quantum computing including qubit, quantum gates, quantum entanglement, etc.. Secondly, we in-depth discuss 5 quantum machine learning algorithms of quantum support vector machine, quantum neural network, quantum k-nearest neighbour, quantum principal component analysis and quantum k-Means algorithm. Thirdly, we conduct discussions on the applications of quantum machine learning in image recognition, drug efficacy prediction and cybersecurity. Finally, we summarise the challenges of quantum machine learning consisting of algorithm design, hardware limitations, data encoding, quantum landscapes, noise and decoherence.

In this article we present a model of adjustable moisture control for historical buildings. Proposed system is developed in a form of flexible IoT infrastructure in which a complex system of sensors is set to measure inside conditions of humidity and compare result to levels of groundwaters, rain and wind speed to manage drying system. Developed control model is using type-2 fuzzy logic reasoning to flexibly adjust decisions to the intensity of water absorption. In this way proposed model makes an innovative intelligent system for control of interior conditions in historical buildings. Developed system was installed and examined in an old brewery building. Research results show efficiency in dehumidification at lowest cost.

The present study proposes a novel method for predicting the discharge capabilities of lithium-ion (Li-ion) batteries using a digital twin model in practice. By combining cutting-edge machine learning techniques, such as AdaBoost and long short-term memory (LSTM) network, with a semiempirical mathematical structure, the digital twin (DT)—a virtual representation that mimics the behavior of actual batteries in real time is constructed. Various metaheuristic optimization methods, such as antlion, grey wolf optimization (GWO), and improved grey wolf optimization (IGWO), are used to adjust hyperparameters in order to optimize the models. As indicators of performance, mean absolute error (MAE) and root-mean-square error (RMSE) are applied to the models after they have undergone extensive training and ten-fold cross-validation. The models are rigorously trained and cross-validated using the NASA battery aging dataset, a widely accepted benchmark dataset for battery research. The IGWO-AdaBoost digital twin model emerges as the standout performer, achieving exceptional accuracy in predicting the discharge capacity. This model demonstrates the lowest mean absolute error (MAE) of 0.01, showcasing its superior precision in estimating discharge capabilities. Additionally, the root mean square error (RMSE) for the IGWO-AdaBoost DT model is also the lowest at 0.01. The findings of this study offer insightful information about the potential utilization of the digital twin model to accurately predict the discharge capacity of batteries.

Heart sound analysis plays an important role in early detecting heart disease. However, manual detection requires doctors with extensive clinical experience, which increases uncertainty for the task, especially in medically underdeveloped areas. This paper proposes a robust neural network structure with an improved attention module for automatic classification of heart sound wave. In the preprocessing stage, noise removal with Butterworth bandpass filter is first adopted, and then heart sound recordings are converted into time-frequency spectrum by short-time Fourier transform (STFT). The model is driven by STFT spectrum. It automatically extracts features through four down sample blocks with different filters. Subsequently, an improved attention module based on Squeeze-and-Excitation module and coordinate attention module is developed for feature fusion. Finally, the neural network will give a category for heart sound waves based on the learned features. The global average pooling layer is adopted for reducing the model's weight and avoiding overfitting, while focal loss is further introduced as the loss function to minimize the data imbalance problem. Validation experiments have been conducted on two publicly available datasets, and the results well demonstrate the effectiveness and advantages of our method.

As The cornerstone of human health, physical and mental well-being are intricately linked, influencing both an individual's physical condition and their emotional state [1]. Chronic diseases such as hypertension and diabetes can have a significant impact on mental health, leading to anxiety and depression [2]. Similarly, psychological problems such as stress, anxiety, and depression can weaken the immune system, making individuals more susceptible to physical illnesses. In recent years, the rapid development of technology has brought exciting new possibilities to the field of physical and psychological health. The Internet of Things (IoT) and artificial intelligence (AI) have shown great potential in building a comprehensive health management system that empowers individuals to take a more proactive role in their well-being.

Digital security in modern systems very often uses biometric, and increasingly, new implementations appear. Such applications can be found everywhere, even when picking up the package from courier, we certify its receipt through our signature on the tablet. However, verification of this form is not one of the simplest elements in information processing systems. Given the different sizes, angles, or writing conditions that may affect its stability, new methods to evaluate signatures are constantly needed. In this article, we propose the use of spline interpolation and two types of artificial neural networks to verify the identity of a person based on selected local and global features extracted from the image of a signature. Global features are extracted concerning interpolation and graphic processing methods, while local features are verified using convolutional neural networks. Both sets of features are used in the identity verification process. The article presents the model of the operation together with experiments, taking into account various parameters of the proposed extraction. We have reached an accuracy of 87.7% on the SVC2004 database.

Abstract Medical and health field is a hot application field of wireless sensor networks. How to correctly refine and classify telemedicine sensor data is the research focus in related fields. Therefore, a detailed classification mathematical model simulation of telemedicine sensor data based on multi feature fusion is proposed. On the basis of telemedicine sensor data acquisition, it is preprocessed to reduce the computational overhead of detailed classification. The reliability features of the preprocessed telemedicine sensing data are extracted, the extracted features are fused by the principal component analysis method, and the refined classification model of telemedicine sensing data is constructed based on the principle of machine learning. The fused features are input into the model to complete the refined classification of telemedicine sensing data. The experimental results show that the correct refinement classification rate of the proposed method is more than 90%, the refinement classification accuracy is higher than 98.5%, the convergence speed is good, and the refinement classification time is 4 ~ 12 s, which proves that the correct refinement classification rate and accuracy of the proposed method are high, the classification time is short, and has good application performance.

Abstract Wireless communication smart bracelet data include motion data, sleep time data, heart rate and blood pressure data and positioning data, etc. These data have diversity and high complexity, and there are interconnections or interactions between the data, which have high clustering difficulty. To this end, a new data clustering algorithm is studied for wireless communication smart bracelets. The K-medoids algorithm is used to calculate the intra-cluster, inter-cluster, or overall similarity to complete the initial clustering of the bracelet data. Setting the clustering evaluation index can determine the optimal number of clusters. The data objects that are closely surrounded and relatively dispersed are selected as the initial clustering centers and combined with the new index IXB to complete the improvement of the data clustering algorithm. The test results show that the accuracy, recall, and F1 of the research algorithm for clustering the heart rate monitoring dataset, temperature monitoring dataset, energy consumption dataset, and sleep monitoring dataset are higher than 97%, which indicates that the data clustering effect of the algorithm is good.

Abstract Modern medical clinics support medical examinations with computer systems which use Computational Intelligence on the way to detect potential health problems in more efficient way. One of the most important applications is evaluation of CT brain scans, where the most precise results come from deep learning approaches. In this article, we propose a novel correlation learning mechanism (CLM) for deep neural network architectures that combines convolutional neural network (CNN) with classic architecture. The support neural network helps CNN to find the most adequate filers for pooling and convolution layers. As a result, the main neural classifier learns faster and reaches higher efficiency. Results show that our CLM model is able to reach about 96% accuracy, and about 95% precision and recall. We have described our proposed mechanism and discussed numerical results to draw conclusions and show future works.

To avoid the overfitting phenomenon that appeared in performing graph neural networks (GNNs) on test examples, the feature encoding scheme of GNNs usually introduces the dropout procedure. However, after learning latent node representations under this scheme, Gaussian noise produced by the dropout operation is inevitably transmitted into the next neighborhood aggregation step, which necessarily hampers the unbiased aggregation ability of GNN models. To address this issue, in this article, we present a novel aggregator, denoising aggregation (DNAG), which utilizes principal component analysis (PCA) to preserve the aggregated real signals from neighboring features and simultaneously filter out the Gaussian noise. The idea is different from using PCA on traditional applications to reduce the feature dimension. We regard PCA as an aggregator to compress the neighboring node features to have better expressive denoising power. We propose new training architectures to simplify the intensive computation of PCA in DNAG. Numerical experiments show the apparent superiority of the proposed DNAG models in gaining more denoising capability and achieving the state of the art for a set of predictive tasks on several graph-structured datasets.

Based on sensor networks within a dynamic and real-time environment, a novel parking-lot navigation method is proposed based on diffusion equation and Poisson equation to achieve convenient and efficient navigation process with edge computing mind to aid information query and navigation. From the perspective of theoretical proof, is presented parallel method mainly for ordinary differential equations (ODEs) by partitioning the time domain. In this article, our new method is combined by parallelization for linear heat equations. The model problem is decoupled into several sub-problems in space-time sub-domains. We prove the super-linear convergence when the time interval is bounded. Numerical experiments testify our theoretical analysis. Simultaneously, the proposed method holds the lower constraint condition and some improper navigation routes can be updated. Mathematical analysis and simulations show that the method is accurately and efficiently enabled to solve typical sensor network configuration information navigation problem.

Advanced models of Artificial Intelligence enable systems of IoT to work with great flexibility to the needs of users. In this article we present our developed IoT system for driving support by the use of type-2 fuzzy logic control module. We have developed the IoT system to collect the data about driving conditions and evaluate them adjusting to the needs of user. Applied module of fuzzy logic of the second type was used in analysis of accelerometers signals to flexibly adjust to uncertainty of evaluation of driving expectations of each driver. Our developed system was tested in different cars by driving on various roads and results show excellent efficiency.

Abstract Existing systems have disadvantages such as slow running speed, long time-consuming, and poor rendering effect in virtual reconstruction of architectural spatial structure. In order to solve such problems, virtual reconstruction system of building space structure is designed using laser 3D scanning technology under condition of fusion of multiple big data. The system was equipped with a 3D laser scanner and connected to computer, and the noise interference was reduced by image preprocessing module to complete the hardware design. The system improved user interface and maintenance module. Eventually, 3D model reconstruction was realized via data acquisition, data registration, coordinate transformation and 3D rendering. The results show that the system designed in this paper runs fast, and color of reconstruction results is consistent, which indicates that reconstruction results of building space structure obtained by the system are conducive to in-depth study of building space.

Abstract Current IoT communication node spacing selection process show may potential areas for improvements such as high delay ratio, high total energy consumption ratio, confusion of the optimal communication information band, intelligent spacing node design under the constraints of the energy-saving selection of IoT communication. Based on energy-saving constraints, the link status between nodes is evaluated through link stability and link quality. In order to prevent the generation of serious noisy nodes and frequency hopping data, the interference nodes under the intrusion of the Internet of Things are identified by determining transition amplitude of the noise nodes in the transmission data sequence. Finally, according to the calculation results of the optimal communication node selection, the design of the intelligent spacing selection model for the communication nodes of the Internet of Things is realized. The simulation results show that the established model not only reduces energy consumption of nodes, shortens the average transmission delay of nodes, but also improves anti-interference effect of node spacing selection.

Abstract In order to promote the efficiency of image retrieval in mobile network and realize the fast query of key images, this paper designs a quick search method of key feature images in mobile networks. The key features of retrieved images are extracted by rotation invariant local binary method. According to the extracted key features of the image, the query target image is processed by coarse quantization then the distance of the key features of the image is calculated and retrieved. Finally, non-exhaustive search method is used to achieve quick search of key feature images in mobile network. Experimental results show that this method can effectively extract specific images, The desired image can be quickly searched by reserved key features, and the F-score value of quick search is higher than 0.9.

Abstract The third type of neural network called spiking is developed due to a more accurate representation of neuronal activity in living organisms. Spiking neural networks have many different parameters that can be difficult to adjust manually to the current classification problem. The analysis and selection of coefficients’ values in the network can be analyzed as an optimization problem. A practical method for automatic selection of them can decrease the time needed to develop such a model. In this paper, we propose the use of a heuristic approach to analyze and select coefficients with the idea of collaborative working. The proposed idea is based on parallel analyzing of different coefficients and choosing the best of them or average ones. This type of optimization problem allows the selection of all variables, which can significantly affect the convergence of the accuracy. Our proposal was tested using network simulators and popular databases to indicate the possibilities of the described approach. Five different heuristic algorithms were tested and the best results were reached by Cuckoo Search Algorithm, Grasshopper Optimization Algorithm, and Polar Bears Algorithm.

Abstract In this article, we want to discuss the use of deep learning model to predict potential vibrations of high-speed trains. In our research, we have tested and developed deep learning model to predict potential vibrations from time series of recorded vibrations during travel. We have tested various training models, different time steps and potential error margins to examine how well we are able to predict situation on the track. Summarizing, in our article we have used the RNN-LSTM neural network model with hyperbolic tangent in hidden layers and rectified linear unit gate at the final layer in order to predict future values from the time series data. Results of our research show the our system is able to predict vibrations with Accuracy of above 99% in series of values forward.

Modifying the existing models of classifiers' operation is primarily aimed at increasing the effectiveness as well as minimizing the training time. An additional advantage is the ability to quickly implement a given solution to the real needs of the market. In this paper, we propose a method that can implement various classifiers using the federated learning concept and taking into account parallelism. Also, an important element is the analysis and selection of the best classifier depending on its reliability found for separated datasets extended by new, augmented samples. The proposed augmentation technique involves image processing techniques, neural architectures, and heuristic methods and improves the operation in federated learning by increasing the role of the server. The proposition has been presented and tested for the fruit image classification problem. The conducted experiments have shown that the described technique can be very useful as an implementation method even in the case of a small database. Obtained results are discussed concerning the advantages and disadvantages in the context of practical application like higher accuracy.

In this article, we propose a model of face detection in risk situations to help rescue teams speed up the search of people who might need help. The proposed lightweight convolutional neural network (CNN) architecture is designed to detect faces of people in mines, avalanches, under water, or other dangerous situations when their face might not be very visible over surrounding background. We have designed a novel light architecture cooperating with the proposed sliding window procedure. The designed model works with maximum simplicity to support mobile devices. An output from processing presents a box on face location in the screen of device. The model was trained by using Adam and tested on various images. Results show that proposed lightweight CNN detects human faces over various textures with accuracy above 99% and precision above 98% what proves the efficiency of our proposed model.

Intelligent Manufacturing 5.0 of multipulse rectifier systems requires them to be optimized for a variety of use in transportation and factories producing hearty touch technology. The research presented in this article show advances of using heuristic models to set 12-pulse and 24-pulse rectifiers to work under low- and high-voltage load. As a result of heuristic optimization electric systems increase efficiency and reduce energy consumption by efficiency benefits in adopting artificial intelligence. Applied heuristic models helped in computer simulations to optimize system settings in a short time. Results show that optimized models are more efficient and our proposed approach is reducing voltage pulsation. As a result optimized system improves electromagnetic compatibility for beneficial use in modern industry and sensible human–machine cooperation.

In this article, we propose malfunction classifications for trucks, a novel idea for smart fleet management systems. In the proposed cooperative cooperative intelligent transportation (C-ITS), the developed neural network work with information from truck fleets to select the trucks that need a service. From the results returned from the deep neural network classifier, the applied heuristic algorithm uses the classification outputs to select the most important results. The proposed process is multithreaded; thus, the composed system gains additional efficiency. The implemented deep learning model achieved an accuracy above 98%, and an above 95% recall. The developed solution was tested on the Scania Truck data collection. The research results show the importance of the advances and validate our concept for potential further development.

In this article, Polar Bear Optimization Algorithm (PBO) is parallelized to solve the problem of computed tomography (CT) with incomplete data. It is vary hard to model correctly 2D and 3D objects by using CT scanners when information is incomplete. Our approach is to use PBO to reduce recovery time and simplify specificity of the phenomenon. Results from our research show that proposed approach is enabling fast and accurate reconstruction of objects modeled in projection space.

The huge popularity of heuristics contributes not only to the improvement and modeling of new solutions but also to their adaptation to selected goals. Recent years have shown the popularity of their use also in machine learning as a training algorithm or allowing for the selection of optimal architecture or hyper-parameters. In this paper, we propose an adaptation of a nature-inspired algorithm for preprocessing images in a parallel way for obtaining higher classification results. The proposed idea is based on analyzing images by heuristic representative which is Red Fox Optimization Algorithm and returning a specific value. These values are used in deciding to classify the entire image or trim it to eliminate unnecessary objects. We modeled this solution and evaluated using the learning transfer method for VOC 2007 dataset. The obtained results were compared on selected classes to show the advantages of a proposal.

The paper proposes a model of regulation for transformer secondary voltages in order to improve quality of rectified voltage. For this purpose, a simulation model for an 18-pulse rectifier system was presented. We have defined optimization task by using proposed parallel version of the Polar Bear Optimization (PBO) algorithm. For the developed method simulation results are presented and discussed. Research show that proposed parallelization enabled efficient optimization of the large number of parameters and, at the same time, shortens simulation time.

The image recognition task is one of the fundamental aspects of image and video analysis. Recognition of individual objects allows for further inference or analysis. Unfortunately, quite often the detection and recognition itself are difficult tasks. Especially if there are many different objects in the image, or if there is some noise. In this paper, we propose a method for extracting specific features from images. The proposition is a hybridization of two main tools - meta-heuristic and fuzzy system. At first, an objective function is created for a specific object, then the meta-heuristic is used for analyzing an image for finding the best features. The operation of creating an objective function and then interpreting the position of individuals in the metaheuristic is evaluated by a fuzzy controller. The use of fuzzy logic enables the creation of decision sets during data analysis. This is possible through the adaptive technique of improving the value of the membership functions in Takagi-Sugeno systems. A fuzzy approach shows great potential in analyzing the position in the image. The proposed feature extraction mechanism has been tested and discussed due to the possibility of using fuzzy logic as well as its hybridization with meta-heuristics.

Technological development increases capacity of information systems, which with development of faster data transfer will be able to host variety of new devices. In this article we present electronic modules, infrastructure and fuzzy rules control model with implemented software for new generation home environment. The system is developed for the next IoT level based on 6G network communication standards. Proposed control model is efficient in water flow management, wind shield control, security aspects and carbon dioxide limitation via adaptive ventilation. Developed infrastructure is ready for new 6G communication standard, which will additionally improve efficiency and data flow at end-user devices and local area level.

Abstract The application of the traditional single frame character image super-resolution reconstruction method has some problems, such as noise can not be removed completely and anti-interference performance is poor. A new method for the super-resolution reconstruction of single frame character image based on wavelet neural network is proposed. The structure and interface of image acquisition unit of solid state image sensor are designed. Combined with pinhole imaging model and camera self-calibration, image acquisition of Internet of Things is completed. An image degradation model was established to simulate the degradation process of ideal high-resolution image to low-resolution image. Wavelet threshold denoising method is used to remove the noise in a single frame character image and improve the anti-interference performance of the method. The wavelet neural network reflection model is used to reconstruct the single frame feature image and improve the resolution of the image. The experimental results show that the blur degree of the reconstructed image is always less than 5%. In the whole experiment, the accuracy of this method can be maintained at 80% ~ 90%. The image detail retention rate of the research method is relatively stable. With the increase of the number of experimental images, the retention rate of image details remains between 80% and 95%, indicating that the method is effective in practical application.

Abstract In view of the problem that the traditional learning service recommendation does not fully consider the distinct differences between individuals, it is easy to lead to the contradiction between unchanging learning resources and learners’ personalized learning needs that are constantly improving, so an adaptive learning service recommendation improvement algorithm based on big data is proposed. Idea is based on adaptive learning platform and function modules. We consider the individual differences between students, to students as the center, collect students’ personalized learning demand data, and according to the data information to build student demand model. On the basis of using data mining methods for clustering recommendation service resources in learning, the adaptive recommend according to students’ individual need is proposed. The experimental results show that the adaptive learning service recommendation algorithm based on big data has high recommendation accuracy, coverage rate and recall rate, which is of great significance in the actual learning service recommendation.

Security of networking in cyber-physical systems is an important feature in recent computing. Information that comes to the network needs preevaluation. Our solution presented in this article is based on deep learning model developed for network traffic analysis of various Internet of things solutions. At the level of firewall or gateway, information about current connection is gathered for the recurrent neural network. The model evaluates this information and forwards decision back to the firewall to take security actions if needed. In the research, we have tested our solution on two open datasets. The results confirm that our model is very efficient in recognition of potential threats reaching above 99% of accuracy even in a case of reduced number of evaluated networking features.

Mixed reality environments give better chances to provide constant help to the people in need. Applied there artificial intelligence models will provide ad hoc monitoring measures, which may be the best chance to protect life in dangerous conditions. In this article, we present our research on mixed reality system developed to detect symptoms of unusual poses at work, home, or other environments. Recurrent neural network is using sensor readings to evaluate the situation by the minimum necessary number of body sensors working as safe indicators. Research results show that the developed system is working with very high accuracy of 99.89% using just two body sensors working in a separate mode. The system can work without any special infrastructure or development in various environments to help workers and elder people in dangerous situations.

The fusion of Internet of Things (IoTs) and computational intelligence makes it possible to increase energetic efficiency of our homes. Connected devices can be optimally adjusted to the needs of a family. In this article, we present our developed IoT convection installation for a small house with the developed remote platform control system. The control module is gartering readings from sensors and information from users about conditions in the house and, by the use of computational intelligence, optimizes parameters to adjust the developed IoT convection system for better comfort of a family. We have done a full convection installation, both in practical and theoretical models, together with remote control system and the proposed security model. Optimization results show increased comfort of use with lower changes in the temperature inside. The system after optimization shows significant improvement in lower changes of the temperature and lower consumption.

Prediction models are widely applied in several fields.In this study we present a discussion on using Recurrent Neural Network as predictor for salaries of future graduates.The model is based on feature analysis which leads to input values of the predictor.We have analyzed several compositions and ideas.As a result we have selected Recurrent Neural Network to be the most accurate.Presented results confirm this selection and show high precision.

The paper proposes a method for classifying and fast retrieving images which uses boosting metalearning to search for the most salient image features. We use local image keypoints as image features. We construct by boosting a set fuzzy rules describing image feature parameters. The rules constitute a set of weak classifiers voting for the final image class. The method can use various image features, engineered and learned by deep learning methods. We checked the methods on some real-world images.

We propose a semi-tensor product attention network model as a visual question answering tool for complex interaction over image features. Proposed model performs matrix multiplication of two arbitrary dimensions, which is used to overcome possible dimensional limitations and improve recognition flexibility. In used block-wise operation we preserve spatial and temporal information but reduce the number of parameters by using low-rank pooling scheme. Applied BERT pre-train model is tuned to recognize question features. The proposed model is evaluated on the VQA2.0 dataset. Research results show that our model has good accuracy and easy reconfiguration for future research.

Abstract The traditional design of voice collector has poor anti attack ability, which makes the encryption effect of voice transmission poor. Therefore, taking the mobile network voice collector as the research object, 3des-ecc algorithm is applied to the information transmission encryption of the mobile network voice collector.An improved speech signal collector is designed, which combines 3DES and ECC algorithm to realize the encryption of speech transmission information. An improved voice signal collector is designed, which combines 3DES and ECC algorithm to realize the encryption of voice transmission information. In the process of encryption, 168-bit random key is generated first, and it is grouped according to 56 bits as 3DES key, and then the plaintext is encrypted by the key to generate ciphertext; the random key is encrypted by ECC public key of the receiver. The experimental results show that the encryption time of this method is less than 1 s, the data integrity is 93%, and the data loss rate is only 0.33%. It has better anti attack ability, fast encryption speed and good encryption effect.

Abstract There are situations when someone finds a certain object or its remains. Particularly the second case is complicated, because having only a part of the element, it is difficult to identify the full object. In the case of archeological excavations, the fragment should be classified in order to know what we are looking at. Unfortunately, such classification may be a difficult task. Hence, it is essential to focus on certain features which define it, and then to classify the complete object. In this paper, we proposed creating a novel soft tree decision structure. The idea is based on soft sets. In addition, we have introduced convolutional networks to the nodes to make decisions based on graphic files. A new archeological item can be photographed and evaluated by the proposed technique. As a result, the object will be classified depending on the amount of information obtained to the appropriate class. If the object cannot be classified, the method will return individual features and possible class.

A new speckle suppression algorithm is proposed for high-resolution synthetic aperture radar (SAR) images. It is based on the nonlocal means (NLM) filter and the modified Aubert and Aujol (AA) model. This method takes the nonlocal Dirichlet function as a linear regularization item, which constructs the weight by measuring the similarity of images. Then, a new despeckling model is introduced by combining the regularization item and the data item of the AA model, and an iterative algorithm is proposed to solve the new model. The experiments show that, compared with the AA model, the proposed model has more effective performance in suppressing speckle; namely, ENL and DCV measures are 21.75% and 4.5% higher, respectively, than for NLM. Moreover, it also has better performance in keeping the edge information.

Internal organs, like lungs, are very often examined by the use of screening methods. For this purpose, we present an evaluation model based on a composition of fuzzy system combined with a neural network. The input image is evaluated by means of custom rules, which use type-1 fuzzy membership functions. The results are forwarded to a neural network for final evaluation. Our model was validated by using X-ray images with lung nodules. The results show the high performances of our approach with sensitivity and specificity reaching almost 95% and 90%, respectively, with an accuracy of 92.56%. The new methodology lowers the computational demands considerably and increases detection performances.

Artificial intelligence methods applied in smart home environments can give a ubiquitous support of people, provide automatic control of system settings to lower the costs of operation, improve energetic efficiency, and longer endurance of components. In this article, we discuss the use of neural networks and rule-based systems as the components of automatic control over house elements. In the proposed framework, we store the knowledge and teach the system about the home in parallel to operation. Initially, the system starts with global settings, however, data are collected during use and parallel to this regular training is run so when the new knowledge guarantees higher efficiency, the system switch to use it. We have developed a new neural-based mechanism with rules control method to lower the costs of operation while keeping the needs of users. The components were tested and discussed due to practical application in our everyday life.

In modern medical systems fast recognition of bacteria strain from microscopy image is very often done by the use of specialized computer systems. For these programs intelligent methodologies are very important tools. In this paper we present a model of bacteria recognition based on a composition of region covariance with Convolutional Neural Networks (CNN). In the first stage an input microscopy image is segmented by the use of region covariance model. Next these segments are forwarded to CNN for recognition of visible bacteria strains. Experiments were done for rod-shaped bacteria and spherical or nearly spherical shape bacteria. The results show high potential of the proposed methodology.

BACKGROUND AND OBJECTIVE: New technology enables constant boost to the powers of mobile devices, which in the previous years have transformed from simple mobile phones to smart phones. Computational powers of these electronics enable actions that previously were possible only for computers. By the use of special applications, we may benefit from sensors and multimedia capabilities of operating systems. Therefore, a new era for devoted implementations opens, in which a smart application can take a role of computing system to estimate the symptoms of diseases by evaluating signals coming from a human body. METHODS: We propose a model of an application implemented for mobile android systems, which can be used for examination of central nervous system motor disorders occurring in patients suffering from Huntington (HD), Alzheimer, or Parkinson diseases. In particular, the model tracks tremors (involuntary movements), and cognitive (memory loss or dementia) impairments using touch and visual stimulus modalities. The proposed model interprets the symptoms from human bodies that indicate one of the diseases of the nervous system. Pre-processing of collected data for feature extraction is executed on a mobile device by using core functionality and methods provided in android's application programming interface. The information is evaluated by a back-propagation neural network classifier and the result is presented to the end user. The system is able to contact medical supervision and provide an assistance from the clinic. RESULTS: The system uses a collected dataset of 1928 records, taken from 11 HD patients and 11 healthy persons in Lithuania, to gather statistics about examinations and presents the results as medical evaluation with prediction on the state of health. The accuracy of recognition of early, prodromal symptoms for central nervous system motor disorders is 86.4% (F-measure 0.859). The app (available on Google Play) is easy to use and is efficient tool for decision support in medical examinations. CONCLUSIONS: The use of intelligent apps which can help to evaluate neurodegenerative disorders is an important enhancement to medical diagnosis. The developed smartphone app supports the doctor with additional results that are easy to compare with other examinations. This kind of examination is a nice change from classic stereotypes, especially for younger age patients, who are used to various aspects of information technology.

Parallel programming is a field of science with a great potential nowadays due to the development of advanced computers architectures.Appropriate usage of this tool can be therefore highly beneficial in multimedia applications and significantly decreases the time of calculations.In this article, we analyze how the speed of calculations is influenced by the usage of parallel algorithms in image filtering processes.We present a method based on multithreading and the division of the image for rectangles.The filter is applied parallel on each part of the image.Results show that in some cases our proposition can bring over 90% benefit when compared to the classical approach.

Today many applications require a high Quality of Service (QoS) to the network, especially for real time applications like VoIP services, video/audio conferences, video surveillance, high definition video transmission, etc. Besides, there are many application scenarios for which it is essential to guarantee high QoS in high speed mobility context using an Internet Mobile access.However, internet mobile networks are not designed to support the real-time data traffic due to many factors such as resource sharing, traffic congestion, radio link, coverage, etc., which affect the Quality of Experience (QoE).In order to improve the QoS in mobility scenarios, the authors propose a new technique named "Smart VPN Bonding" which is based on aggregation of two or more internet mobile accesses and is able to provide a higher end-to-end available bandwidth due to an adaptive load balancing algorithm.In this paper, in order to dynamically establish the correct load balancing weights of the smart VPN bonder, a neural network approach to predict the main Key Performance Indicators (KPIs) values in a determinate geographical point is proposed.

Multimedia devices are widely used in professional applications as well as personal purposes.The use of computer vision systems enables detection and extraction of important features exposed in images.However constantly increasing demand for this type of video with high quality requires simple however reliable methods.The objective of presented research is to investigate applicability of heuristic method for real-time video frames capturing and correction.

We propose a novel noise cancellation method based on the scale-adaptive remixing and demixing of Intrinsic Mode Functions (IMFs) constructed using Empirical Mode Decomposition (EMD). The method addresses the problem of mode mixing in the EMD by performing mode demixing. An illustrative example using noisy random binary sequence is presented. The proposed approach allows achieving better denoising results than the classic first IMF discarding approach.

Nowadays the effective and fast detection of fruit defects is one of the main concerns for fruit selling companies.This paper presents a new approach that classifies fruit surface defects in color and texture using Radial Basis Probabilistic Neural Networks (RBPNN).The texture and gray features of defect area are extracted by computing a gray level co-occurrence matrix and then defect areas are classified by the applied RBPNN solution.

This work presents a new adaptive technique for image compression based on Discrete Wavelet Transform (DWT) and Radial Basis Function Neural Networks (RBFNN). The technique can be employed both for lossless and lossy (higher) compression and has been devised in order to deal effectively with a large variety of images. Proposed solution performs well both in terms of computing time and memory. Its generality, flexibility and efficiency make it attractive for storage and transmission in the field of vision and multimedia systems.

In this paper a novel approach to automatic medical signal diagnosis is proposed. The authors propose a solution based on the application of Computational Intelligence (CI) to assist classification procedure based on Neural Network (NN) attempt. Experiments have been performed with the proposed CI solution applied on retrieved digital signals of various heart actions to help in missing or incomplete data gathering. Then this knowledge was applied as a training set in proposed NN classifier used to recognize potential dangers and therefore work as Decision Support System (DSS) designed for medical purposes.