Metaverses embedded in our lives create virtual experiences inside of the physical world. Moving towards metaverses in aircraft maintenance, mixed reality (MR) creates enormous opportunities for the interaction with virtual airplanes (digital twin) that deliver a near-real experience, keeping physical distancing during pandemics. 3D twins of modern machines exported to MR can be easily manipulated, shared, and updated, which creates colossal benefits for aviation colleges who still exploit retired models for practicing. Therefore, we propose mixed reality education and training of aircraft maintenance for Boeing 737 in smart glasses, enhanced with a deep learning speech interaction module for trainee engineers to control virtual assets and workflow using speech commands, enabling them to operate with both hands. With the use of the convolutional neural network (CNN) architecture for audio features and learning and classification parts for commands and language identification, the speech module handles intermixed requests in English and Korean languages, giving corresponding feedback. Evaluation with test data showed high accuracy of prediction, having on average 95.7% and 99.6% on the F1-Score metric for command and language prediction, respectively. The proposed speech interaction module in the aircraft maintenance metaverse further improved education and training, giving intuitive and efficient control over the operation, enhancing interaction with virtual objects in mixed reality.

Towards Aircraft Maintenance Metaverse Using Speech Interactions with Virtual Objects in Mixed Reality

Image segmentation is a branch of digital image processing which has numerous applications in the field of analysis of images, augmented reality, machine vision, and many more. The field of medical image analysis is growing and the segmentation of the organs, diseases, or abnormalities in medical images has become demanding. The segmentation of medical images helps in checking the growth of disease like tumour, controlling the dosage of medicine, and dosage of exposure to radiations. Medical image segmentation is really a challenging task due to the various artefacts present in the images. Recently, deep neural models have shown application in various image segmentation tasks. This significant growth is due to the achievements and high performance of the deep learning strategies. This work presents a review of the literature in the field of medical image segmentation employing deep convolutional neural networks. The paper examines the various widely used medical image datasets, the different metrics used for evaluating the segmentation tasks, and performances of different CNN based networks. In comparison to the existing review and survey papers, the present work also discusses the various challenges in the field of segmentation of medical images and different state-of-the-art solutions available in the literature.

/pdf/deep-neural-networks-for-medical-image-segmentation-t5dkc3ri.pdf

Deep Neural Networks for Medical Image Segmentation

Distributing the system workload and balancing all incoming requests among all processing nodes in cloud computing environments is one of the important challenges in today cloud computing world. Many load balancing algorithms and approaches have been proposed for distributed and cloud computing systems. In addition the broker policy for distributing the workload among different datacenters in a cloud environment is one of the important factors for improving the system performance. In this paper we present an analytical comparison for the combinations of VM load balancing algorithms and different broker policies. We evaluate these approaches by simulating on CloudAnalyst simulator and the final results are presented based on different parameters. The results of this research specify the best possible combinations.

Performance evaluation and analysis of load balancing algorithms in cloud computing environments

Reinforcement and deep reinforcement learning for wireless Internet of Things: A survey

Power restoring time in power distribution systems (PDS) can be minimized by using efficient fault localization techniques. This paper proposes a novel, robust and scalable cloud based internet of things (IoT) solution for identification and localization of faults in PDS. For this purpose, a new algorithm is developed that can detect single and multiple simultaneous faults in the presence of single and multiple device or sensor failures. The algorithm has utilized a zone based approach that divides a PDS into different zones. A current sensing device (CSD) was deployed at the boundary of a zone. The function of CSD is to provide time synchronized current measurements and communicate with a cloud server through an edge device (ED). Another contribution of this research work is the unique implementation of context aware policy (CAP) in ED. Due to CAP, only those measurements are transmitted to cloud server that differ from the previously transmitted measurements. The cloud server performed calculations at regular intervals to detect faults in PDS. A relational database model was utilized to log various fault events that occur in PDS. An IEEE 37 node test feeder was selected as PDS to observe the performance of our solution. Two test cases were designed to simulate individual and multiple simultaneous faults in PDS. A third test case was implemented to demonstrate the robustness and scalability of proposed solution to detect multiple simultaneous faults in PDS when single and multiple sensor failures were encountered. It was observed that the new algorithm successfully localized the faults for all the three cases. Consequently, significant reductions were noticed in the amount of data that was sent to the cloud server. In the end, a comparison study of a proposed solution was performed with existing methods to further highlight the benefits of our technique.

Cloud Based IoT Solution for Fault Detection and Localization in Power Distribution Systems

In recent years, the growth of Internet of Things (IoT) as an emerging technology has been unbelievable. The number of networkenabled devices in IoT domains is increasing dramatically, leading to the massive production of electronic data. These data contain valuable information which can be used in various areas, such as science, industry, business and even social life. To extract and analyze this information and make IoT systems smart, the only choice is entering artificial intelligence (AI) world and leveraging the power of machine learning and deep learning techniques. This paper evaluates the performance of 11 popular machine and deep learning algorithms for classification task using six IoT-related datasets. These algorithms are compared according to several performance evaluation metrics including precision, recall, f1-score, accuracy, execution time, ROC-AUC score and confusion matrix. A specific experiment is also conducted to assess the convergence speed of developed models. The comprehensive experiments indicated that, considering all performance metrics, Random Forests performed better than other machine learning models, while among deep learning models, ANN and CNN achieved more interesting results.

Performance Analysis and Comparison of Machine and Deep Learning Algorithms for IoT Data Classification.

Cloud computing users are faced with a wide variety of services to choose from. Consequently, a number of cloud service brokers (CSBs) have emerged to help users in their service selection process. This paper reviews the recent approaches that have been introduced and used for cloud service brokerage and discusses their challenges accordingly. We propose a set of attributes for a CSB to be considered effective. Different CSBs’ approaches are classified as either single service or multiple service models. The CSBs are then assessed, analyzed, and compared with respect to the proposed set of attributes. Based on our studies, CSBs with multiple service models that support more of the proposed effective CSB attributes have wider application in cloud computing environments.

Cloud service selection using cloud service brokers: approaches and challenges

Numbers of various programming models have been proposed to process big data in recent years. However, MapReduce is the most famous programming model amongst cloud computing environments and includes many advantages, yet there are several challenges to deal with. Load balancing is considered as one of the most significant downsides of MapReduce which causes the increase in applications' runtime and accordingly results in less-efficiency, where there is no appropriate proposed mechanism. Although, data locality and data skew are known as two main key factors for determination of load balancing, yet it is remarkable that load balance highly depends on whether the computational clusters are homogeneous or heterogeneous. This paper examines the effectiveness of two main key factors. These are data locality and data skew on homogeneous and heterogeneous clusters. Besides, a review is conducted on a number of recent literature in the same context of load balancing improvements in Hadoop MapReduce. Finally, all investigated researches are compared with the purpose of highlighting the differences of various load balancing methods, the optimisation phase, type of clusters and the main challenges.

Load balancing in MapReduce on homogeneous and heterogeneous clusters: an in-depth review

Unobtrusive and smart recognition of human activities using smartphones inertial sensors is an interesting topic in the field of artificial intelligence acquired tremendous popularity among researchers, especially in recent years. A considerable challenge that needs more attention is the real-time detection of physical activities, since for many real-world applications such as health monitoring and elderly care, it is required to recognize users' activities immediately to prevent severe damages to individuals' wellness. In this paper, we propose a human activity recognition (HAR) approach for the online prediction of physical movements, benefiting from the capabilities of incremental learning algorithms. We develop a HAR system containing monitoring software and a mobile application that collects accelerometer and gyroscope data and send them to a remote server via the Internet for classification and recognition operations. Six incremental learning algorithms are employed and evaluated in this work and compared with several batch learning algorithms commonly used for developing offline HAR systems. The Final results indicated that considering all performance evaluation metrics, Incremental K-Nearest Neighbors and Incremental Naive Bayesian outperformed other algorithms, exceeding a recognition accuracy of 95% in real-time.

Meysam Vakili

Papers

Performance Analysis and Comparison of Machine and Deep Learning Algorithms for IoT Data Classification.

Cloud service selection using cloud service brokers: approaches and challenges

Load balancing in MapReduce on homogeneous and heterogeneous clusters: an in-depth review

Incremental Learning Techniques for Online Human Activity Recognition.