Mobile Networks and Applications最新文献

In remote teaching, massive resource data types have heterogeneous diversity attributes. Currently, recommendation algorithms only consider the optimal solution in the local domain under an attention mechanism to ensure efficiency, without considering the embedding correlation of recommendation features in the entire local domain, resulting in suboptimal recommendation results in a massive data environment. This paper proposes an improved multi-objective intelligent recommendation algorithm for massive remote teaching resources. The logical framework of a multi-objective intelligent recommendation algorithm for massive resources is provided. First, connections between different domains are constructed through knowledge graphs as well as global domain embedding are generated related to users and remote teaching resources. Then, recommendation representations of users and teaching resources in the target domain are expressed through fully localized embedding representations. Finally, the recommendation representation is trained through the output layer to output the target domain recommendation prediction score for remote teaching resources. The average and diversity of remote teaching resource prediction scores are used as evaluation parameters for the recommendation list, and a multi-objective optimization algorithm is adopted to optimize the calculation process of recommendation prediction scores through operations such as crossover and mutation of initial solutions. A new prediction score of remote teaching resource recommendation is generated and compared with existing methods to obtain a better recommendation list. Experimental results show that the MRR values of the recommended results of this method are all above 0.985, and the MAE value is controlled below 0.5. The recommended results are accurate and can effectively improve the teaching performance of students in different majors, improve prediction scores, diversity scores, and satisfaction.

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.

The secure transmission of information is receiving more and more attention nowadays. Aeolus is a novel platform designed to enhance the development of distributed applications by preventing unauthorized disclosure of information. And one of the most representative systems for information transmission is the file system, therefore it is of great significance to formally analyze the Aeolus-based file system. In this paper, we use Communicating Sequential Processes (CSP) to model and formalize the file system based on Aeolus. Moreover, we utilize the Process Analysis Toolkit (PAT) to simulate and verify the CSP description of our established model. We specifically verify the validity of five properties: Deadlock Freedom, Divergence Freedom, Reachability, Secrecy, and Integrity. The verification results demonstrate that the model successfully satisfies these properties, affirming the effectiveness of the framework in ensuring file operations and guaranteeing the secure transmission of information.

Colorectal cancer poses a global health risk, often heralded by colorectal polyps. Colonoscopy is the primary modality for polyp detection, with precise, real-time segmentation being key to effective diagnosis and surgical planning. Existing segmentation models like convolutional neural networks (CNNs) and Transformers have propelled progress but face trade-offs between precision and speed. CNNs excel in local feature extraction yet struggle with global context, while Transformers handle global information well but at a computational cost. Addressing these constraints, we introduce TMPSformer, a groundbreaking lightweight model tailored for efficient and accurate real-time polyp segmentation. TMPSformer, with its compact size of only 2.7 M, features a pioneering hybrid encoder merging Transformers’ long-range dependencies and shift Multi-Layer Perceptrons (MLPs)’ local dependencies, effectively enhancing segmentation performance. It also equips an All-MLP decoder to streamline feature fusion and enhance decoding efficiency. TMPSformer utilizes the Flash Efficient Attention (FEA) module to replace the traditional Attention module, significantly improving real-time performance. A comprehensive evaluation on five public polyp segmentation datasets demonstrated TMPSformer’s superiority over existing state-of-the-art algorithms. Specifically, TMPSformer achieves real-time processing at 162 frames per second (FPS) at 512 × 512 resolution on the Kvasir-SEG dataset using a single NVIDIA RTX 2080 Ti GPU, and achieves a mean Intersection over Union (mIoU) of 0.811. Its segmentation performance surpasses ColonSegNet by 8.7% and SegFormer by 4.8%. Additionally, TMPSformer significantly reduces complexity, cutting the parameter count by 1.8× and 31× compared to ColonSegNet and SegFormer, respectively.

Mobile Medical information systems MMIS or mHealth applications support personal health and potentially improve the health sector by offering a solution to significant problems faced by the healthcare system. While espousing these Mobile Medical Information Systems, sequestration and security issues arise. Due to advanced computing and different capabilities, data security and confidentiality come major enterprises with continuously expanding mHealth operations. European Union General Data Protection (GDPR) and California Consumer Sequestration Act (CCPA) raise mindfulness; still, they need to address developing a system that meets sequestration and security conditions. This paper deals with research literature to understand current privacy and security issues and possible solutions for patients and providers using mHealth applications. This is the reason for several threats, such as information harvesting, tracking patients, relaying attacks, and denial of service attacks, which affect the confidentiality and integrity of these devices. We discussed the challenges and risks associated with Mobile Medical information systems and emphasized the need to address these concerns for widespread adoption. Mitigation strategies include robust security measures, regulatory compliance, and user awareness. We discussed the impact of privacy and security issues on healthcare, including potential harm to patients and disruptions in system functioning, reviewing laws, conducting a literature review, and assessing mHealth system applications. We emphasize the need for comprehensive security measures and continuous evaluation of security practices in mHealth, which need to be addressed to achieve quality, continuity, and portability of health services. We offer a critical and methodical assessment of the state of the art in mHealth security and privacy and suggest a methodology for creating and executing MMIS that is safe and protects privacy.

Sports rehabilitation focuses on the restoration of physical function and performance of martial arts students and athletes by assisting them in the recovery process during injuries. Each athlete’s injury is unique and requires personalized treatment. The conventional approaches lack tailored feedback and precise monitoring to provide personalized treatment, depending on the nature of an injury. To enhance treatment outcomes in sports rehabilitation, this paper utilizes an improved Recurrent Neural Network (RNN) model that is optimized for sequential data analysis and incorporates attention mechanisms to prioritize relevant features from profiles of marital art students in colleges, injury details, and rehabilitation protocols. It uses wearable devices of the Internet of Things (IoT) to collect sequential data from different sources in real-time. Next, the gathered data is cleansed and preprocessed, which ensures compatibility with temporal data structures and facilitates seamless integration into clinical settings. This process includes different techniques like normalization, segmentation, and feature extraction. Finally, an RNN model is reconfigured, which consists of the input layer, two hidden LSTM layers, and an output layer that facilitates the processed data of the athletes. The athlete’s progress is continuously monitored, and timely adjustments are made to rehabilitation plans. The model is then trained on diverse datasets, which include athlete profiles, injury characteristics, rehabilitation protocols, and outcome measures. Experimental results demonstrate a 15% increase in prediction accuracy and a 20% improvement in rehabilitation efficiency. Additionally, player performance metrics showed a 25% enhancement in recovery speed and a 30% reduction in the risk of re-injury.

Musical education has a positive impact on psychological health. It enhances emotional expression and contributes to constructive transformation of mental health. This study explores the use of a machine learning technique known as Q-learning to assess these effects. The research process commences by collecting data from music students. This data includes psychological health status, emotional expression levels and progress in musical education. Surveys and regular assessments are used for this purpose in which Students report their psychological health and emotional experiences. It also tracks and record their progress in musical education. Secondly, a Q-learning algorithm is implemented to analyze the collected data. It demonstrates how different musical education activities influence psychological health and emotional expression. The algorithm works in the form of iterations and can learn from interactions and make decisions based on rewards. Thirdly, the algorithm processes the information and identifies which activities have the most positive impact on musical education by identifying patterns. It also assists in suggesting different types of improvements and methods in teaching methods. To evaluate the performance of the study different performance metrics are used. These indicators include psychological health scores, levels of emotional expression, progress in music skills, attendance rates, participation in class activities and student engagement levels. It also depicts what kinds of activities are particularly beneficial in increasing impact of the musical education. The study shows that students deeply engaged in music have better psychological health and exhibit higher levels of emotional expression.

In recent years, the development of wearable devices and health applications has influenced the technical development of SHM in sports-related activities. These technologies can be invoked to improve the health management of college students who practice certain physical activities. This paper proposed and developed a novel IoT framework for sports health monitoring using prediction models based on big data analytics and convolutional neural networks (CNN). The proposed framework combines IoT technology with state-of-the-art deep learning techniques to analyze extensive data collected from wearable devices, optimizing sports performance and mitigating injury risks. The study outlines a complete methodology, including data collection from multiple sources, preprocessing for CNN models, and constructing and comparing CNN-based predictive models. Experimental results reveal the effectiveness of the proposed technique in predicting injuries and optimizing performance results. Ethical considerations, such as data privacy, model interpretability, and fairness, are also discussed to ensure responsible implementation. The findings highlight the potential of CNN and big data analytics in enhancing sports health management, offering personalized recommendations, and promoting overall well-being among college students. The experiment results outperformed the performance of the different evaluation metrics such as accuracy, sensitivity, specificity, F1 score, and MCC, with the proposed model achieving 0.9342%, 0.8500%, 0.9415%, 0.8803%, and 0.8232%, respectively. The error losses achieved less than those of the other methods, such as MSE, MASE, MAE, and RMSE, which achieved 0.0654%, 0.0758%, 0.2356%, and 0.2537%, respectively. Future research should focus on refining the models, expanding the dataset, and addressing ethical concerns to improve the framework’s applicability and effectiveness further.

This study examines the impact of musical therapy on psychotherapy conditions such as stress in college students. College students encounter numerous stressors including academic pressure and social challenges. It negatively impacts their physical and mental well-being which leads to anxiety and depression. Musical therapy has been recognized as a tool for stress reduction. However, the mechanisms underlying its effectiveness remain unclear. Therefore, this research utilizes Mixed Density Neural Networks (MDNN) to analyze the physiological responses associated with musical therapy. The initial phase focuses on collecting multi-modal data both with and without music. This data is collected from college students using surveys and physiological sensors. In the second phase data is preprocessed to remove noise and anomalies which is then followed by feature extraction which captures relevant information from the signals. In the third phase, the collected data is analyzed using MDNN, capable of handling both continuous and categorical data. It has an input layer, two hidden layers, a mixed-density layer, and an output layer. The input layer uses a linear activation function to process data from physiological sensors and musical stimuli features. The first and second hidden layer uses ReLU activation functions and has 50 and 25 neurons respectively. The mixed-density layer has one neuron and uses a sigmoid activation function for adaptive connection density based on input data. Finally, the output layer has one neuron and a linear activation function to predict stress levels. Evaluation metrics such as accuracy, precision, recall, and F1-score are used to assess the performance of the model. It shows that slow and calming music significantly reduces stress levels among college students. Moreover, the implementation of the proposed algorithm improved the accuracy of stress level predictions by 20% and outperformed its predecessors.

The integration of Internet of Things (IoT) technology in healthcare plays a significant role in remote health management. It enables real-time data collection and patient monitoring. This research study aims to enhance data accuracy, reliability, and predictive capabilities of the IoT network in healthcare by exploring advanced covariance techniques, which include Kalman filters, particle filters, and covariance intersection. Kalman filters process real-time data by minimizing the mean of the squared error and estimating the state of a system accurately. Particle filters are used to handle non-linear systems and provide accurate estimates using a set of random samples, while Covariance intersection fuses data from multiple sources. It does this without needing any knowledge of the correlation between various variables, which makes it ideal for IoT applications. Initially, data is collected from wearable sensors, home monitoring systems, and mobile health applications. Wearable sensors measure heart rate, blood pressure, and glucose levels. Home monitoring systems track environmental factors and patient activities, and Mobile health applications gather patient-reported data. Secondly, Data preprocessing techniques are used to clean the data and handle missing values. Kalman filters provide continuous health updates. Particle filters predict health trends, and Covariance intersection integrates data from multiple IoT devices. To evaluate the performance of these covariance techniques compared with traditional schemes such as simple averaging, weighted averaging, and basic linear regression using various performance metrics, which include Mean Squared Error (MSE), Root Mean Squared Error (RMSE), correlation coefficients, Precision, Recall, F1 Score and Area Under the Curve (AUC). The results show that covariance methods have enhanced overall system performance by 20% in terms of accuracy, 15% in precision, and 18% in recall. By fusing data seamlessly, covariance intersection ensures an accurate understanding of patient health across different environmental and situational contexts.