Face super-resolution generates high-resolution face images from low-resolution inputs, supporting face recognition in challenging environments. While deep learning methods, like Stable Diffusion... Show More

Face super-resolution generates high-resolution face images from low-resolution inputs, supporting face recognition in challenging environments. While deep learning methods, like Stable Diffusion and origin Transformer-based framework, have advanced super-resolution, they require heavy computation, making them difficult for tasks like face recognition. However, recognition accuracy only needs images of size 112x112, reducing the necessity for extremely large outputs. Existing methods often rely on convolutional attention, limiting receptive fields and performance. To address this, we propose SETFSR, a Semantically Guided Efficient Attention Transformer Face Super-Resolution. Our model leverages efficient self-attention for global feature extraction and incorporates face parsing constraints for structural accuracy. Experiments on Helen, CelebA, and FFHQ datasets show that SETFSR outperforms state-of-the-art models in PSNR, SSIM, and identity preservation.

In modern industrial production, trajectory tracking control of multi-degree-of-freedom robotic arms entails safety and efficiency. This study focuses on a two-degree-of-freedom manipulator... Show More

In modern industrial production, trajectory tracking control of multi-degree-of-freedom robotic arms entails safety and efficiency. This study focuses on a two-degree-of-freedom manipulator, developing a mechanical dynamics model and implementing a parallel Linear Active Disturbance Rejection Controller (LADRC) for joint control. To enhance robustness and responsiveness, a hybrid adaptive control method is introduced, integrating adaptive laws (AL) for online tuning of parameters and offline optimization using the Deep Deterministic Policy Gradient (DDPG) algorithm. This improves tracking control efficiency and effectiveness. Additionally, offline reinforcement learning refines parameters without adding real-time computational demands. Finally, semantic web principles ensure interpretability and transparency of the control flow. Compared to traditional LADRC, the proposed method shows superior anti-interference capabilities and dynamic performance.

The static control limits restrict self-healing capabilities and decision-making processes, impeding adaptability to the dynamic shifts in intricate industrial operations, frequently leading to... Show More

The static control limits restrict self-healing capabilities and decision-making processes, impeding adaptability to the dynamic shifts in intricate industrial operations, frequently leading to suboptimal or anomalous conditions that undermine production efficiency. This paper presents a methodology for the identification of suboptimal operating conditions with respect to yield and quality. A GSR model for the identification of suboptimal operating conditions of yield and quality based on random forest classification was established. The experimental results demonstrate that the method is capable of rapidly and accurately identifying the production and quality of the calcining kiln. The identification accuracy of the yield and quality of the suboptimal operating conditions is 99.82% and 99.18% respectively. In the production process, real-time identification of operating parameters enables rapid detection of suboptimal operating conditions in yield and quality, providing the basis for optimal regulation and control, which in turn can improve production efficiency.

A brain tumor is one of the most prevalent causes of cancer death. The best strategy is the timely treatment of brain tumors in their early detection. Magnetic Resonance Imaging (MRI) is a standard... Show More

A brain tumor is one of the most prevalent causes of cancer death. The best strategy is the timely treatment of brain tumors in their early detection. Magnetic Resonance Imaging (MRI) is a standard non-invasive method to detect brain tumors. For early detection and better patient survival through MRI scans, the diagnosis needs a high level of knowledge in the radiological and neurological domains to identify the cancers. Researchers have suggested various brain cancer detection techniques. However, most existing automatic cancer detection approaches suffer from poor accuracy and low detection rates. This paper proposes a hybrid deep learning (DL) using deep feature extraction and adaptive Mask Region-based Convolutional Neural Networks (Mask-RCNNs) model for brain tumor detection and classification method to overcome these issues. The experimental findings on the benchmark dataset demonstrate that the planned model is highly effective, with 99.64% accuracy, 95.93% precision, 95.39% recall, and 95.67% F1-score.

With the growth of web technology, the semantic web offers a promising framework for online knowledge collaboration. However, trust issues can undermine users' willingness to collaborate, reduce the... Show More

With the growth of web technology, the semantic web offers a promising framework for online knowledge collaboration. However, trust issues can undermine users' willingness to collaborate, reduce the frequency of interaction and collaboration efficiency. This paper introduces a super node-based trust management model designed to enhance semantic networks by linking nodes through trust relationships. The model exploits the synergistic incentives of similar interest behaviours to achieve a steady construction of trust relationships. We propose a similarity filtering algorithm that calculates the similarity to filter out false, misleading, or unfair information effectively. Through simulations, we compare our model with RRGRET, Surework, and community-based approaches, and the results show that our model has good network properties, while also resisting multiple malicious attacks and guaranteeing collaboration success. This research contributes to optimizing node relationships within semantic networks and strengthening network robustness against interference.

At present, the medical field requires the development of clinical diagnostic and therapeutic robots. The research team successfully transformed the biophysical structures of mitochondria and lipid... Show More

At present, the medical field requires the development of clinical diagnostic and therapeutic robots. The research team successfully transformed the biophysical structures of mitochondria and lipid particles from 2D to 3D. Based on cloud point registration algorithm and SVM, LSSVM, and Bayesian algorithms, the predicted cloud point was added to the cloud point volume to increase the cloud point data capacity and combined with psychological analysis of 3D cloud point structure hierarchical clustering. The K-SVD algorithm is used to deconvolve pathological data to reduce color interference on visual transformers, fit 3D data of midbrain pathways, and finally fuse 2D data converted from pathological data, mandala data, and immune factor data with brain computer interface data for classification prediction, improving the robustness of artificial intelligence diagnostic algorithms. This information fusion robot can accurately diagnose the comprehensive condition of patients and contribute to medical assisted diagnosis.

Stock Market Prediction (SMP) has developed into a significant area of research, especially in recent decades. Major novelty of the work is to develop an Evolutionary Bidirectional Long Short-Term... Show More

Stock Market Prediction (SMP) has developed into a significant area of research, especially in recent decades. Major novelty of the work is to develop an Evolutionary Bidirectional Long Short-Term Memory (EBi-LSTM) framework depends on investors' sentiment in tweets to Stock Market (SM). In addition, three feature selectors: the Chi-Square Test (CST), Analysis Of VAriance (ANOVA) technique and Mutual Information (MI) method are introduced for the selecting most important features. Levy Flight Fuzzy Social Spider Optimization (LFFSSO) algorithm is used for optimal tuning of parameters in the Bi-LSTM classifier. EBi-LSTM algorithm has been worked on datasets like Twitter, Stock, Weather, and Coronavirus disease (COVID-19). The proposed model extends the Valence Aware Dictionary and sEntiment Reasoner (VADER), TextBlob, and robustly optimized Bidirectional Encoder Representations from Transformers Retraining Approach (RoBERTa) for sentiment analysis. Highest results of 88.26%, 90.43%, 89.33% and 92.63% for precision, recall, F1-score and accuracy has been attained by proposed system.

Conventional methods that merge multiple images with different exposure levels often suffer from blur and ghosting due to object movement. Existing ghosting removal algorithms are usually complex... Show More

Conventional methods that merge multiple images with different exposure levels often suffer from blur and ghosting due to object movement. Existing ghosting removal algorithms are usually complex and slow, making them unsuitable for real-time video applications. To address this challenge, this study proposes a novel ghosting-free image fusion method using a dual gain video stream setup on an FPGA. IMX662 image sensor is employed, which simultaneously captures both HCG and LCG images with the same exposure time, enabling efficient HDR image synthesis. The proposed method directly addresses the source of the problem, eliminating the need for post-processing steps, thereby preserving algorithmic simplicity. Experimental results reveal that the proposed method not only removes ghosting by 100% but also processes data on an FPGA 98.79% faster than traditional software-based HDR fusion techniques, enabling real-time video stream processing. This dual gain, ghosting-free fusion algorithm demonstrates promising potential for use in high-speed photography and surveillance.

As the tourism, car rental and car-sharing market rapidly expand, after the recovery from the epidemic, consumer demand for car rental services is increasing. When many consumers need to rent a car... Show More

As the tourism, car rental and car-sharing market rapidly expand, after the recovery from the epidemic, consumer demand for car rental services is increasing. When many consumers need to rent a car, they hope to use it conveniently and quickly. They also need to book parking spaces along the way and have corresponding insurance services. However, there is a lack of effective integration between the existing car rental system and reserved parking spaces. As a result, users often need to spend extra time and energy looking for parking spaces when using car rental services, and the insurance policies generated by temporary insurance services are difficult to cover at the same time. To address these issues, this study proposed an integrated car rental service platform based on blockchain technology and the Interplanetary File System. This platform consolidates transaction information among various stakeholders in the car rental service. It also preserves the consumer's complete rental order information and status, uses the Elliptic Curve Diffie-Hellman to privately generate a shared key.

Generative AI has revolutionized the field of digital art by enabling automated image creation that mimics human artistry. Traditional approaches often need more thematic coherence and help to... Show More

Generative AI has revolutionized the field of digital art by enabling automated image creation that mimics human artistry. Traditional approaches often need more thematic coherence and help to capture complex styles in artistic domains. In order to overcome these constraints, we present a Generative Adversarial Network (GAN) model integrated with a Semantic Web architecture, therefore providing automated painting creation for creative businesses. Our strategy guarantees thematic and stylistic congruence by using Semantic Web methods to include contextual information and GANs to create high-quality paintings from noise. Our GAN produces reasonable and varied creative outputs. Experimental findings show how flexible the model is for different artistic topics, therefore helping applications in the creative sectors via improved AI-driven artistic production.

Recently, financial institutions have increasingly focused on risk control to protect assets and maintain trust. However, the sensitivity and privacy of data make data sharing for analysis... Show More

Recently, financial institutions have increasingly focused on risk control to protect assets and maintain trust. However, the sensitivity and privacy of data make data sharing for analysis challenging. Current systems typically rely on local data, making it difficult to handle cross-institutional data or complex fraudulent activities. To address this issue, we propose a financial risk control model based on the Federated Semantic Web (FSW). This model deploys a federated learning framework between banks, allowing data to be stored independently, thus ensuring privacy protection. The model automatically selects and scores risk features from transaction data using a multi-perspective topological graph, and assigns the final risk labels via a weighted voting mechanism. The central model updates and redistributes the global model based on parameters shared by local models through the federated averaging algorithm. Experimental results show that the proposed method effectively identifies fraudulent transactions and predicts credit risk, meets security standards, and improves accuracy.

Addressing these challenges requires innovative monitoring techniques that can accurately assess the impact on land use and land cover. To address these issues, this study introduces a Semantic... Show More

Addressing these challenges requires innovative monitoring techniques that can accurately assess the impact on land use and land cover. To address these issues, this study introduces a Semantic Web-enabled framework combined with superpixel-based image segmentation for effective environmental monitoring. Leveraging aerial imagery and artificial intelligence to monitor these effects, employing a Simple Linear Iterative Clustering (SLIC) algorithm for image segmentation and a combination of classifiers (Support Vector Machine, SVM; Random Forest, RF; and Naive Bayes, NB) for land change detection. The effectiveness of this Semantic Web-enabled approach is demonstrated by its ability to accurately identify areas affected by ASGM, with robust statistical metrics including a kappa coefficient of 0.7616, an F1 score of 0.8806, and a Jaccard index of 0.798. These results underscore the method's capability in providing detailed insights into land cover changes, thereby serving as a significant tool for environmental monitoring and aiding in policy formulation.

Campus guide systems are crucial to university infrastructure, shaping the experiences of students, staff, and visitors. Current systems face critical challenges in three areas: capturing diverse... Show More

Campus guide systems are crucial to university infrastructure, shaping the experiences of students, staff, and visitors. Current systems face critical challenges in three areas: capturing diverse user needs, translating emotional requirements into design elements, and integrating campus cultural identity. This study integrates Kansei Engineering (KE) and Generative Artificial Intelligence (AIGC) to propose a semantic-driven design method. Using Semantic Web and Natural Language Processing (NLP), it models demand semantics, extracts emotional semantics such as safety and belonging, and maps them to design semantics for AIGC to generate personalized guide solutions. The approach leverages data-driven emotional semantic analysis and generative models to improve path guidance precision and cultural representation. Results indicate significant improvements in user experience, pathfinding accuracy, and cultural communication, with higher user satisfaction. This method provides a new semantic-driven pathway for developing campus guide systems and development prospects.

Researchers in human motion prediction have focused on mathematical modeling of the human skeletal structure, often overlooking the spatio-temporal characteristics of human pose sequences. To... Show More

Researchers in human motion prediction have focused on mathematical modeling of the human skeletal structure, often overlooking the spatio-temporal characteristics of human pose sequences. To address this, we propose the multi-scale spatio-temporal information aggregation net (MSTIA-Net), which includes two key modules: the graph convolutional spatio-temporal information aggregation (GCSTIA) module and the windowed discrete cosine transform (WDCT) temporal encoding module. GCSTIA extracts and integrates multi-scale temporal and spatial features of human motion sequences, while WDCT removes high-frequency noise and compresses data. model's efficacy is demonstrated on three datasets: Human3.6, CMU, and 3DPW, achieving performance improvements of 2.4%, 4.1%, and 1.7%, respectively.

In multi-UAV-assisted mobile edge computing (MEC), insufficient consideration of collaborative computation in inter-UAV communication can significantly reduce computational service capabilities. For... Show More

In multi-UAV-assisted mobile edge computing (MEC), insufficient consideration of collaborative computation in inter-UAV communication can significantly reduce computational service capabilities. For this problem, we present a multi-UAV-assisted MEC offloading optimization model that jointly optimizes task offloading decision, UAV resource allocation, UAV trajectories and establish collaborative computation through inter-UAV communication. First, to solve the multi-UAV-assisted MEC offloading optimization issue, we define a weighted utility function that balances delay and energy consumption. Then, to tackle the continuous nature of the computation-offloading problem and the coexistence of discrete and continuous variables, the PPO algorithm is enhanced by integrating an average reward objective function and a hybrid action generation offloading mechanism. Finally, we propose a multi-UAV-assisted MEC computing offloading optimization method to improve the utility function. Experiments show that the proposed method significantly enhances system utility.

The hazards associated with cyber security attacks have increased due to our increasing reliance on digital technologies. We suggest an ensemble strategy employing multiple models of machine... Show More

The hazards associated with cyber security attacks have increased due to our increasing reliance on digital technologies. We suggest an ensemble strategy employing multiple models of machine learning to get over these restrictions. Our suggested approach detects an extra 141 dangerous URLs, outperforming the single model that is currently in use by 6%. We gathered a collection of malicious and benign URLs for this investigation. Applying Principal Component Analysis (PCA) to feature extraction. Subsequently advantageous features are selected and categorised using the Social Spider optimised Densely-connected Convolutional Caps net (SSO-DCN). The most effective detection frameworks and automated methods for identifying rogue web pages, including those distribute malicious malware, are included in the proposed technology. When comparing the proposed method to the existing approaches, the improvements are in mean accuracy (96.30%), precision (96.90%), F1-Measure (97.20%), recall (96.90%), FPR (3.50%) and FNR (3.90%).

Semantic Web technologies in education provide opportunities to improve learning outcomes and organizational efficiency via organized and significant information representation. However, with... Show More

Semantic Web technologies in education provide opportunities to improve learning outcomes and organizational efficiency via organized and significant information representation. However, with progress in educational technology, there is an absence of frameworks using Semantic Web capabilities to provide tailored learning experiences and enhance administrative procedures efficiently. Hence, this study proposes a framework called “Education using Semantic Web Tools (E-SWT),” employing ontologies, metadata tagging, and intelligent reasoning systems. The method supports seamless platform interoperability, adaptive content delivery, and efficient resource management. The experimental results demonstrate that the proposed model increases the learning experience ratio by 96.12%, Organizational Efficiency ratio by 97.53%, Fostering Collaboration by 98.82%, Data Management ratio by 97.68%, and Personalized Learning Paths by 96.66% compared to other existing models.

Advanced persistent threats (APT) attacks are very low-frequency, hidden in network traffic, and it's difficult to detect APT attacks with traditional detection methods. Therefore, this paper... Show More

Advanced persistent threats (APT) attacks are very low-frequency, hidden in network traffic, and it's difficult to detect APT attacks with traditional detection methods. Therefore, this paper presents a Semantic Web-driven unsupervised detection framework for APT attacks in network traffic, leveraging Enhanced DBSCAN++ and aiNet algorithms. The proposed DBSCAN++ based aiNet clustering algorithm (D-aiNet) is firstly used for clustering the training samples on the extracted network traffic, and then generates a Self Set (Ss) and an Antibody Set (As). In the testing phase, an Immune Network based anomaly detection model (INAD) is employed to match network traffic samples from the selected dataset with antibodies of the Antibody Set (As) in order to detect APT attacks. To validate the effectiveness of UADDN, experimental studies were conducted using the SCVIC-APT-2021 dataset. Compared with UADAIN, and the other detection methods based on K-means clustering, aiNet-HC clustering, the experimental results demonstrate that the proposed UADDN shows superior performance in detecting APT attacks.