Virtual Reality (VR) is an immersive simulation of the environment. The quality of tracking in VR applications is measured in terms of accuracy and robustness. Tracking based on sensing the environment using only one type of sensor may not be sufficient to meet the requirements of VR applications. Hybrid tracking based on inputs from two or more sensors can cope with the challenging needs of such applications. This paper discusses the implementation of two mobile-based VR applications designed for Military shooting training. The first application is developed using a single type of tracking technique named Virtual Shooting Tutor (VST) without hybrid tracking. In the second application, the inputs from external Micro Electro-Mechanical Systems (MEMS) gyroscope and those from mobile integrated gyroscope are hybridized. The application is named VST with hybrid tracking. Two versions of VST are compared with each other to evaluate the impact of hybrid tracking on the accuracy and robustness of the VR application.

The growth of the 4th industrial revolution (Industry 4.0) and the improved implementation of Big Data are motivating the need for data-directed and decision-making in manufacturing. Digital Twin (DT) is one of the top data-directed and decision-making concepts allowing businesses and manufacturers to simulate products to build faster, cost-effective, and high-quality products. The motorized (Automotive / Vehicle) Industry is estimated to hold 15% of digital twin use cases last year. DT technology has contributed from the starting stage of design to the final construction stages of vehicles and also helps the driver of the vehicle to draw useful information from its daily functions and makes the driver more comfortable and safer.

Deep learning (DL) is a cutting-edge image-processing technology that includes various satellite image sources being processed to analyze, enhance, and classify. This article covers a multilayer DL framework that classifies different types of vegetation and land cover using IRS p6 satellite images from many time scales and sources. The core of the design is an ensemble of supervised NNs and unsupervised neural networks (NNs) for optical imaging categorization and incomplete data restitution due to mists, reflections, and other natural effects that affect images. In this article, we contrast the traditional densely integrated multilayer perceptron (MLP) with the most popular method in remote sensing field random forest as the basic supervised NN architecture with convolutional NNs (CNNs). In general, utilizing the aforementioned procedure reduced accuracy and required longer computation times to train the model which produced 94.3%. The hyperparameters to adjust are the number of neurons, input layer, optimizer, number of epochs, filter size, and iterations. The second stage involves adjusting the number of layers. Some other conventional algorithms are lacking in this. The accuracy might be impacted by many layers. To overcome that, applying the Adam optimizer will produce a higher accuracy level of 96.72% with faster computation time and less memory management.

In recent years, artificial intelligence applications in education have expanded daily. The study and implementation knowledge of this is necessary for building advanced applications based on AI that help in the education sector. The first objective of this research paper is to identify the various factors that help predict student performance related to background, education, and psychology. The second objective is to study and analyze the various data mining classifiers (Decision Tree, Naive Bayes, Logistic Regression, SVM, Random Forest, and KNN) and artificial neural networks on the dataset, including 689 records of students for the prediction of student performance in Python programming language. The conclusion of the evaluation metric of all algorithms suggests that ANN implementation is best for classifying student performance as either high or low. It predicts 96% accuracy.

Due to recent developments in technologies like the Internet of Things, Cloud Computing, Fog Computing and Mobile computing, a large volume of data is generated, accessed remotely from various devices like mobile devices, laptops, personal computers, etc. Cloud computing is characterized by a large volume of inbound and outbound traffic, which is critical. This data needs to be protected from intruders while it is in motion andat rest. A Network Intrusion Detection System (NIDS) is a tool that detects network intrusions and raises an alarm if any malicious packet is found on the network. Many researchers have used machine learning algorithms to develop NIDS models to detect intrusions. Recently, researchers have proposed deep learning techniques to increase the accuracy of NIDS. However, it has been observed that these techniques still fall short of the desired accuracy. In this paper, we have proposed a Deep Support Vector Machine (D-SVM) model with a hybrid feature selection method to increase the accuracy. We believe that no single kernel type can give the best classification results. Hence, we combined multiple SVMs with different kernel types in series to increase accuracy. We trained and tested the model on the NSL-KDD dataset. We learningdataset the same dataset to compare our model with the deep learning model. We observed that the proposed model outperformed the deep learning model. On reaching our model with the previous work, we observed that the proposed model gives better results than earlier models, which used SVM and Deep learning techniques. It was found that the proposed model has better accuracy andperforms well on other performance metrics.

One of the significant contributors to the present death rate for women is breast cancer. An earlier diagnosis of this illness can save costs and increase survival rates. In the past, machine learning algorithms that were used to categorize data were only dependent on manually created features, which could not adequately capture variations and needed a higher degree of classification accuracy. The classification problem was also solved using deep learning methods. The experiment seeks to categorize the cancerous cell types — benign and malignant - using a convolutional neural network (CNN) based model. It is the most incredible technique for classifying images since it is one of the best at identifying the items and patterns in an image. It is a that have a kind of deep learning approach that is best known for spotting patterns in images; it aims to present images in an abstract form that contains the most soundless information required for differentiating them from other images with a similar appearance. BreakHis, a dataset with microscopic biopsy images, and BUSI, a dataset with histopathology images, were employed for this investigation. The primary goal of this experiment is to experiment with the model used to diagnose breast cancer by categorizing various breast photos as either cancerous (Malignant) or non-cancerous (Benign).

Detection of Credit Card Fraud becomes the primitive source of concentration. Investigators would follow a trend of calling a cardholder and invoke the respective discussion with a conclusion labeled as “genuine” or as “fraudulent” to alert the status of the transaction and update the same information to the respective person involved. Examined the effectiveness of the Novel Random Forest with Convolutional Neural Network (CNN) and eXtreme Gradient Boosting (XGBoost) methods. Both techniques have shown advantages in various facets of fraud detection. The innovative Random Forest algorithm excelled in recognizing anomalies and innovative fraud patterns, while the CNN model showed its capacity to capture complex temporal patterns within transaction data. The fraud detection of credit card transactions is analyzed here using machine learning algorithms like Convolutional Neural Network (CNN), Novel Random Forest (RF) and eXtreme Gradient Boost Algorithm optimized with the sparrow search algorithm. The pretest powers had been carried out with 80% for training and 20% for testing with the Kaggle data set. A statistically significant difference in three algorithms with two-tailed values would be p=0.001(p<0.05) in statistical analysis. In Prediction of Credit Card Fraud, the Novel Random Forest has performed better Credit Card Fraud detection than CNN and XGBoost. The Proposed Novel RF algorithm obtained 82.65%, the XGBoost obtained 77.98% and the CNN algorithm obtained 74.98% accuracy.