Intense electromagnetic phenomena occurring on the Sun’s surface give rise to solar flares. Energetic solar flares have the potential to reach Earth, causing significant interference with telecommunication systems. Particularly powerful solar events can even disrupt satellite and ground communication infrastructure, posing a substantial risk of extensive damage. To mitigate these risks, comprehensive monitoring systems and robust forecasting techniques are essential for early detection and warning. This research proposes an approach utilizing the extreme gradient boosting (XGBoost) algorithm for solar flare classification. Recognizing that the performance of XGBoost is heavily influenced by appropriate hyperparameter selection, a modified metaheuristic algorithm is introduced to optimize the network’s hyperparameters. To evaluate the efficacy of the proposed methodology, a real-world dataset is utilized, and a thorough comparative analysis is conducted, encompassing several contemporary algorithms that address the same solar flare classification task under identical conditions. The objective is to identify the advantages and strengths of the proposed modified metaheuristic approach in comparison to existing methods.
IT governance (ITG) is the processes that ensure the effective and efficient use of IT in enabling an organization to achieve its goals. ITG ensures the organization, its data, and its people are protected, while reducing risk and syncing with business objectives. This is achieved through the appropriate IT policies, processes, and procedures, that must be applied consistently across the organization. This paper presents the way forward for adopting the ITG. It introduces the IT Security Office proposal that helps any Libyan organization or institution planning to deploy IT Governance. The proposal is based on the need for any organization to secure; according to the international IT policy and standards, (1) the current use, operation, and implementation of its Information Technology (IT) systems, and (2) the vision of the organization to adopt for example electronic administrative and financial systems and starting its digital transformation journey. This document is organized as follows. First, the proposal introduces IT and its Security Office function and adoption. Then, the proposal specifies the administrative hierarchy of the office in the organization chart and why we need it. Third, it defines the working conditions of the office in such an organizational chart. Fourth, the office duties and responsibilities are described. Finally, the proposal gives recommendations of how an organization could start working in this office and highlights some benefits of implementing such approach.
Many outage probability events have recently been recorded in Libya because of rain attenuation impact, especially at millimetre wave (mmWave) spectrum in terrestrial and satellite links. Therefore, accurate prediction of rain attenuation phenomena is highly recommended to mitigate outages due to rain fade. Rainfall statistical data measured for long-term data of 30 years were collected from five locations in Libya. Collected data were analysed and processed to convert into one–minute rain rate cumulative distribution. The model proposed by ITU-R has been utilised to predict and investigate rain fade based on converted 1-min rain rate data for sites in various climate regions in Libya in the zenith paths of the Hylas 2 satellite. The impact of different parameters such as frequency (GHz), satellite elevation (degrees), polarisation, and percentage of the year on rain fade prediction has been tested. Additionally, the essential properties of link performance have been investigated in determining dependable fade margin. The impact of rain attenuation at 28 GHz was considered for performance analysis at Benina (R0.01
The Holy Qur'an is the oldest comprehensive Arabic book of recommendation for Muslims worldwide. Provide knowledge and information used in various ways. People often utilize the Holy Qur'an, considered a reliable and trustful legislated text, for education and to meet their needs and inquiries of the Muslim community. Like extracting an answer span from the provided passage, The Qur'an has the potential to captivate the curiosity of non-Muslims and propel them towards exploring a vast array of topics and pursuing answers. Over the past few years, Question Answering (Q.A.) has drawn much attention from the NLP community. Researchers and experts have developed various Qur'anic Question Answering (QAA) systems. Nevertheless, the main challenge in the Arabic language is the need for more resources, making it difficult to provide highly accurate Arabic QA systems. The first Qur’an Question Answering shared task workshop, “Qur’an QA 2022,” aims to promote state-of-the-art research on Qur'anic question answering QA in general and machine reading comprehension MRC in particular. It aims to develop models to extract questions answering the holy Qur’an passages. This research paper motivated by this task, suggests an ensemble learning model based on Arabic-supported versions of BERT, which will be implemented using the KNIME platform. We aim to use this model for Arabic Question Answering.; as a result, we get 0.488 for the AUC and 0.946 for accuracy.
This paper presents the development of an advanced system designed to efficiently process Arabic queries in today’s data-driven landscape. By using the T5 sequence-to-sequence model developed by Google and fine-tuning it specifically for transforming textual and verbal input into SQL queries, the system becomes adept at accurately comprehending and interpreting Arabic queries. The resulting system serves as a user-friendly tool that automates the generation of SQL queries. Users can input their queries in written or recorded Arabic utterances, eliminating the need for manual translation and query construction. This involved fine-tuning T5 models on a SQL dataset, splitting the dataset, tokenizing it, and setting training parameters. The implementation phase included loading the fine-tuned model, which incorporated PICARD to generate valid queries effectively. This paper also explores the impact of Arabic translation on the performance and accuracy of the model. The optimal testing on test set accuracy achieved was 63.11
The primordial part of realizing a computer system is conceptual modelling, which has met with great success in relational database design using an ER model. Unfortunately, with the development of the computer field, relational databases confronted shortcomings in their applications and the object-relational model became used more and more. The tools of presenting conceptual modelling do not support all new concepts introduced by such models and the syntax of the SQL4 language. This paper aims to survey the previous works and various proposals on how to create rules for designing the object-relational (O-R) database to indicate their strengths and limitations. The paper provides a comprehensive explanation and evaluation of the existing approaches and the possibility of transferring all data semantics and relationships. This work demonstrates the importance and difficulty of these guidelines and discusses many previous works in this field. This paper confirms that modern technologies and new databases play an important role in the success of software design.
Decoupling the control plane from the forwarding plane is suggested by Software Defined Networks (SDNs). The OpenFlow SDN control protocol can be used to deploy SDN for virtual switches communications in this new architecture, which provides centralized management and monitoring. The creation of SDN technology based on software programming necessitates tests and measuring equipment that can assess performance. However, the open-source performance tools for SDN architecture that are currently available only test the fundamental attributes, despite the fact that there are numerous SDN controllers that require evaluation with relation to topology discovery time. As a performance monitoring tool for SDN controllers, this study suggests the Topology Discovery Tool (TDT). This tool can measure how quickly various topology sizes converge. The northbound interface of Python programming was used to create this utility. The tool is used in the research to evaluate the performance of well-known SDN controllers in systems with various topology sizes and out-of-band control strategies, including an Open Network Operating System, OpenDaylight, and Floodlight.
In this study, we propose a methodology for classifying electrocardiogram (ECG) signals into normal and myocardial infarction (MI) classes. The methodology consists of three main steps: pre-processing, segmentation, and classification. ECG signals obtained from the PTB database are initially subjected to pre-processing using the Daubechies wavelet transform to filter out noise and enhance signal quality. Subsequently, the signals are segmented into 651-sample segments and further reduced to 500 samples per segment for dimensionality reduction. These segmented signals serve as input data for a convolutional neural network (CNN) model, which extracts relevant features and performs the classification task. The proposed methodology achieves an impressive classification accuracy of 97.8
Software refactoring modifies the internal structure of the software without changing its original functionality. Several studies used different machine learning classification algorithms to predict software refactoring opportunities for developers. However, the issue of an imbalanced dataset still needs to be addressed. An imbalanced dataset is a scenario that arises when we have unequal distribution of classes in a dataset. One way to balance a dataset is by adding instances to the minority class (Oversampling) or deleting instances from the majority class (Undersampling). In this paper, four popular techniques of both categories have been evaluated for their capability of improving the imbalanced ratio of highly imbalanced datasets belonging to software refactoring. We used eight publicly available datasets from Java open-source software systems. Four classification algorithms, Naive Bayes, SVM, Forest Trees, and Decision Tree, were applied to evaluate the results of each resampling technique. The experimental results show that oversampling techniques performed better than random undersampling. In particular, when the datasets were balanced with the synthetic minority oversampling technique (SMOTE), the random forest ML algorithm obtained higher scores in different evaluation metrics on all datasets.
This paper investigates signal attenuation in the mobile user interface of a 4G cellular network and proposes an AI-driven strategy to enhance path loss prediction. Employing the Particle Swarm Optimization (PSO) algorithm coupled with AI techniques, it analyzes and forecasts path losses. Measurements were conducted at 1800 MHz and 2100 MHz frequencies in a suburban area of Tebessa. The objective was to minimize the Root Mean Square Error (RMSE) between theoretical and observed path losses by optimizing parameters for the COST231 and Ericsson models using AI-supported PSO. The optimized models, benefiting from the synergy between AI and PSO, exhibited superior performance compared to unoptimized empirical models.
Dumb and hearing-impaired people are unable to communicate as well as normal people; thus, they must rely on sign language. Which is a visual or gestural form of communication. Unfortunately, sign language is neither common nor easy to learn. Consequently, deaf-mute people encounter many challenges in their daily communication. Hence, we proposed an intelligent system that employs a vision-based approach dedicated to Arabic sign language recognition (ArSLR). The system is aimed to recognize Arabic words expressed in dynamic sign language expressions and translate them into textural form. While maintaining natural and flexible translation, the system doesn’t impose any hardware requirements, colored gloves, or limitations on the background. A custom dataset is utilized in the development process of the system. A significant amount of experimental work has been conducted to come up with effective and generalized algorithms for image processing, feature extraction, feature selection, and practical classification. Satisfactory results are achieved with the use of linear discriminant analysis (LDA) as a feature selection and dimensionality reduction method; where 100
Air pollution poses the world’s most significant environmental health risk to people everywhere. According to certain World Health Organization (WHO) estimates, air pollution exposure results in millions of premature deaths each year, with children under five making up a significant portion of these fatalities. While it is estimated that more than 90
Selected bandgap voltage reference circuits based on bipolar transistors from the literature are implemented and simulated using the same process node. This allowed us to compare their performance parameters. The process node used for the simulation is 65 nm PTM which is publicly available. Out of the four designs selected, design II shows the best overall results with temperature coefficient of 17.3 ppm/ ºc, supply voltage ranges from 0.7 to 1.21 V, PSRR of -120 @ 10 Hz, and reasonable gate area.
A fractional order PID (FOPID) controller was designed to control a pressurized water reactor (PWR) nuclear power plant. The reactor was represented by a simplified model using point kinetic equations with one group of delayed neutrons and the associated thermal hydraulic equation. The FOPID design was based on minimizing the sum of integral of squared error (ISE) and integral of squared controller output (ISCO) using particle swarm optimization (PSO) algorithm. Compared with other optimization methods, the PSO is faster, simpler and needs fewer parameters to be adjusted. The simulation of the reactor power response to a reactivity insertion using the FOPID controller showed better performance with respect to the power increase overshoot compared to those of the PID controller. In addition, the settling and steady state times were shorter.
Lung diseases, such as viral pneumonia, lung cancer, and the coronavirus, have caused significant deaths worldwide in the past year. The detection of such diseases by radiologists is a challenging task. However, by applying image processing and artificial intelligence techniques, the computer can help radiologists identify chest diseases more accurately. Our research work aims to propose a methodology to detect chest diseases using convolutional Neural Network (CNN) and image processing techniques more effectively. In this paper, two predefined CNN model architectures (VGG-16, and VGG-19) are applied with Contrast Limited Adaptive Histogram Equalization (CLAHE) as a pre-processing technique in order to enhance the chest X-ray images. Moreover, Gradient Weighted Class Activation Maps (Grad-CAM) were applied to localize the affected areas in chest X-ray images. The two models are trained on the COVID-19 Radiography Dataset provided by Kaggle. For the performance evaluation, the two CNN models are investigated and evaluated with the image-enhanced technique. The experimental work demonstrates a significant improvement in performance using the CLAHE technique for both models. The results show that the enhanced VGG-16 model achieved a 97
Phishing is a popular cyber-attack that tricks users into revealing confidential data such as login passwords or account details, on a website that appears official. Many methods for identifying phishing attacks, particularly the use of machine learning, have been presented in recent years. This study examines the impact of using the Recursive Feature Elimination technique on the classification accuracy of phishing websites by analyzing several Machine Learning algorithms for identifying phishing websites based on URL attributes. According to the experimental findings, RFE (Recursive Feature Elimination) was essential in reducing the effect of irrelevant features on the accuracy of various classifiers. We also assessed the key features included in both datasets, which are crucial for classifying URL-based phishing websites. Our research uncovered four essential features: Page Rank, Links in Tags, Statical_Report and Having_Sub_Domain.
Event abstraction is a process of extracting main events from a large set of data, allowing investigators to identify patterns, connections, and anomalies in event logs that may reveal further evidence of malicious activity. In this paper, we investigate the use of event abstraction in a forensic timeline. This work applies the Drain method, a tree-based abstraction approach, and demonstrates its efficiency in producing accurate event abstraction. It also discusses the challenges faced by investigators in event abstraction and its analysis in a forensic timeline. Finally, this paper presents case studies of web server attacks and creates their event abstraction from a forensic timeline.
Arabic dialects are commonly used on social media platforms by Arabic speakers to express their opinions and connect with each other. However, due to the lack of standardized rules or grammars, analyzing Arabic dialects with NLP tools can be more challenging than standard Arabic. Moreover, the poems domain within Arabic dialects is considered to be more challenging than other domains due to structural differences between poems and regular expressions. This study investigates the use of TF-IDF and N-grams with Lemmatization techniques to develop machine learning classifiers for sentiment analysis in the domain of poems within the Libyan dialect, specifically the Libyan Middle Region dialect. Three experiments were conducted using ML classifiers, namely, SVM, NB, and LR. The first experiment explored classifiers’ performance using .TF-IDF with Unigrams, whereas the second experiment investigated the use of TF-IDF with Trigrams, and the third experiment examined the impact of combining Unigrams and Trigrams on the classifiers’ performance. The experimental results indicate that utilizing Unigrams with TF-IDF can enhance classifier performance, whereas Trigrams with TF-IDF can have a negative effect on the classifiers’ performance. Notably, SVM achieved the highest accuracy of 69.04
Birds are crucial for maintaining ecosystem balance and detecting environmental threats. Therefore, monitoring bird populations is essential for understanding ecosystem health. However, many birds are isolated in high-elevation habitats, making it challenging for researchers to study and monitor their populations. As a result, detecting birds based on their sounds can provide a passive, low-labor approach to monitoring bird populations. This research aims to develop a deep-learning model that accepts an audio waveform of arbitrary length and then acoustically recognizes the species. Toward this end, a subset of the BirdCLEF 2022 dataset was used. This dataset covers 152 bird species. However, four species were selected based on their ecological significance and data availability, allowing comparison and evaluation of different strategies and techniques. The developed model achieved an 89.66
Online phishing is one of the most serious cyber-attacks, with the goal of fraudulently obtaining sensitive information from internet users. Site and Common Resource Locator (URL) phishing occurs when attackers use phishing websites that appear identical to legitimate websites to trick end users into disclosing personal information. Passwords and credit card numbers are examples of sensitive factors. Several techniques have been proposed to address this issue. Machine learning algorithms have proven to be extremely effective in this field. This paper describes a method for detecting phishing websites that is based on the Naive Bayes classifier (GA) and the genetic algorithm (GA). These algorithms are combined to form a wrapped feature selection algorithm, which searches the feature space for a useful and relevant feature. This will then aid in determining whether the sites are genuine or fraudulent. To discover the optimal subset, GA was used to generate feature subsets and the NB classifier was used to evaluate the classification accuracy of feature subsets generated with GA. The proposed method has been tested on three phishing site datasets. On all data sets, experimental results show that NB combined with GA can achieve accuracy of up to 91
