Umm Al-Qura University
To vote for the Best Poster Award, please use this form: (https://uq.sa/ztMwlP) or scan the following QR code:
You can use the above form more than once; however, only your latest vote will be considered.
Supervisory Control and Data Acquisition (SCADA) systems are essential part of industrial control system (ICS) that make up critical infrastructure, such as power supply generator, nuclear reactor, oil and gas industries, water plants, and transportation systems. Because SCADA systems are integrated in an infrastructure that mostly support the basic needs of preserving a human life, the security and protection of SCADA systems is utmost important. Moreover, cyber attacking SCADA systems can be a target in cyberwars due the resulting huge damage with less costs. The main contribution of this paper is to illuminate the possible security measures at early stages that usually neither organizations nor research community focus on. The proposed security measures are identified based on analyzing the top 10 recent attacks on SCADA systems. This paper aims to close the gap between security measures and real-world threads.
Keywords: Control and Data Acquisition (SCADA), Cyber-attacks, Physical Security, Cybersecurity
 |
Email the Presenter |
Providing students with the expertise of scientific and practical skills in Web development requires good preparation and practical competencies, the formation of positive attitudes toward the given course and determining suitable learning outcomes is one of the most important trends and issues in instructional design and technology. One of the most important challenges in this paper is to develop the relatively old course plan to cope with the rapid developments in the world of web application development, this requires Curriculum mapping to identify and address academic gaps, and redundancies. For purposes of improving the overall coherence of the course, its relationship with the other prerequisite courses should be explained more, some learning outcomes may be achieved through more than one course, therefore this study traced other related courses to the problem statement, such as Object-Oriented-Programming (OOP), Software Modelling and Design and other courses. This interconnected set of courses was motivated by the idea of Crosscutting concepts (CCCs) in Web Application Development, such as design patterns, Model-View-Controller (MVC), Entity Framework, Object Relational Mapping, Application Programming Interface (API), and others. This paper proposed some best practices and guidelines to take students gradually from a simple to more complicated tasks in learning Web development.
Keywords: Modeling; Web Development; Crosscutting Concepts
 |
View Poster |
 |
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Cryptocurrency has become a significant business stream in recent years for investors worldwide. Through a giant decentralized bitcoin network, cryptocurrency solutions have been extended to world investors over the internet. Understanding the underlying infrastructure of the bitcoin network is essential for internet users to realize the business environment for their investment and for researchers to improve the provisioning of cryptocurrency solutions. This paper provides a literature review to illustrate the primary methodology of bitcoin networks, underlying infrastructure, and the significant technical challenges and security issues. We describe the blockchain network and its components and structure.We also discuss the bitcoin transaction handling over the blockchain networks and the cryptographic techniques for securing the blockchain networks and cryptocurrency solutions.
Keywords: Cryptocurrency, bitcoin, blockchain network, cryptography.
|
Email the Presenter |
Abstract— Depression is a mental illness that affects a person’s feelings and causes them to be negative. According to the WHO (World Health Organization), 280 million people suffer from depression, and it is a main cause of suicide and carries a great burden of disease. Given that social media is the number one source nowadays for expressing a person’s emotions and feelings, it provides a proper environment to harvest raw data and detect signs of depression by analyzing the content shared by users. In this paper, we proposed a model for the detection of depression using Twitter as a source of information as it is one of the most popular social media platforms. Moreover, we found that research on such topics is lacking for the Arabic language and Arab users even though Arabic is the 4th most used language on Twitter. Therefore, we aimed to detect depression by identifying depressive tweets using natural language processing (NLP) tools and techniques to reveal the sentiments expressed by Arabs in their tweets and the polarity of depression. We have applied machine learning algorithms: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Logistic Regression (LR) and Naïve Bayes (NB) on tweets that are of the Saudi Dialect of the Modern Arabic Language to identify depressive tweets. Along with a combined Term Frequency - Inverse Document Frequency (TF-IDF) and N-gram feature extraction approach, we concluded that combining TF-IDF with N-gram produces better results. We also found that Logistic Regression outperformed the other algorithms with an accuracy of 82%.
Keywords: Sentiment Analysis, Arabic Language, Natural Language Processing, Machine Learning
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
 |
Play Poster Presentation Video |
IoT has been around for a few years now and it’s only gaining more momentum as technology evolves. There’s no doubt that the internet has played a big role in its growth. Nowadays, a lot of inanimate objects can communicate with each other sans human intervention. For example, there are smart devices like smartphones, home appliances, vehicles, and even flight management systems. IoT devices are often unsecured, leaving them open to attack by unauthorized persons. These individuals can intercept data or manipulate the devices for their own malicious purposes. This paper reviews the security threats that face the Internet of Things. It does this by reviewing several related surveys on the security of IoT. In doing this, the paper will discuss the Cryptographic Schemes in IoT. It will also discuss the various light weigh Cryptography algorithms and the design differences for normal block ciphers
Keywords: Cryptographic Algorithms, Internet of Things(IoT)
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
|
Play Poster Presentation Video |
In recent years, there has been a growing interest in the field of Big Data thanks to its benefits, such as cost optimization, improved efficiency, and potential risks identification. Big data has become a widely exploited research area in academia and industry as it benefits both organizations and consumers. However, like any new technology, there are still several challenges relevant to this field and most notably data security. The security challenges of individual’s sensitive information associated with Big Data have been widely studied in the literature. In addition, current studies of Big Data mainly focus on how to minimize the privacy risk brought by Big Data operations; however, unwanted disclosure of the user’s sensitive information may also happen in the process of data collection and analysis. In this paper, we provide a comprehensive study of the security and privacy issues related to Big Data from a wider perspective and investigate various mechanisms that can be used to protect sensitive information and thus, achieve better data security.
Keywords: Big data, privacy, security, sensitive information
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Research into Arabic Aspect Category Detection (ACD), a subtask of Aspect-Based Sentiment Analysis (ABSA) is primarily focused on experimenting with Machine Learning (ML) and Deep Learning (DL) models, but rarely consider methods for robust dataset creation or augmentation. Therefore, this study investigates and compares two applications of AraBERT vr2 on the Arabic SemEval2016 hotel dataset, where the objective is to predict aspect categories in the reviews. The baseline pipeline uses the AraBERT classification layer to predict 34 aspect categories, and the proposed pipeline performs a data augmentation process prior to classification, where the HARD dataset is used with a Word2Vec model to extend the original training of the SemEval2016 dataset. Both implementations of AraBERT were evaluated on Gold Standard dataset derived from SemEval2016. Results indicate that both applications achieved a closely similar f1-score of 0.663 for the baseline and 0.661 for the augmented model. The by-class results are further discussed.
Keywords: Aspect Category Detection, ACD, AraBERT, Aspect-based Sentiment Analysis (ABSA), Text Augmentation, Data Augmentation (DA)
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
|
Play Poster Presentation Video |
There are many customer service skills which every employee needs to master if they are to deal with customers on a regular basis. Without these skills, a company's customer service department can be thrown into chaos and disappointments, not to mention the loss of customers if the support service continues to let them down. In order to facilitate the customer with appropriate support, many applications exist and contribute in different ways to provide the speed of service and break down barriers. This is especially important in large spaces crowded with goods, reducing the time and effort for customers to reach the parts they desire to purchase or acquire is essential. The researchers of this paper contacted employees working in a home supplies store and many questions were raised about the difficulties they face if the store is overcrowded with customers who need the same services simultaneously. Moreover, several personal interviews with customers were conducted to obtain sufficient information, confirm the problem, and develop proposed solutions that are feasible and beneficial to both parties. The result of the data collection and analysis suggested the need to develop a mobile application to help customers receive better services and organize the multiple requests from employees, thus facilitating better communication. It also intends to help employees provide better service, reduce quality service problems and misunderstandings, as well as reduce the waiting time for customers when checking for and finding items.
Keywords: Mobile Application; Customer Service; Service Management
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
|
Play Poster Presentation Video |
With the expansion and improvement of technology and communication in recent years, the necessity to secure and defend these communications from various threats has increased. In this research, we provide a novel method for encrypting audio signals utilizing Baker and discrete cosine transform (DCT) chaotic maps. Three typical audio samples of funky, muted trumpet, and flute are converted into 2D to enable the use of chaotic maps. Then, a DCT is used to extract the coefficients, and the Baker chaotic map is used to replace them. Finally, an inverse DCT is used to perform the encryption. The DCT is applied to the sample during the decryption stage, and the coefficients are altered using the Baker map. An inverse DCT is then applied, and the output is transformed from 2D to 1D to obtain the original audio signal. The effectiveness of the proposed coding model based on Baker maps and DCT is evaluated using a variety of measures, including entropy, histogram, correlation coefficient, spectrogram, noise effect, and differential tests, and the results demonstrate the effectiveness of the proposed encryption model.
Keywords: audio, Baker, chaotic, DCT, encryption
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Given the growing interest in security systems, researchers and developers are continually devising new methods of using biometrics, in particular fingerprints. However, hackers have discovered a flaw in the fingerprint method and exploit it through presentation attacks (PA). As a result, there is a need to protect the safety and security of authentication systems against PA through automatic presentation attack detection (PAD). The current research showed promising results after the application of PAD to deep learning (DL) algorithms. The proposed model is a software-based approach, namely a binary classification system to distinguish between live and fake fingerprint images as these were the most appropriate in terms of cost. In order to achieve the goal of this research, we applied a three-layer convolutional neural network (CNN) using the ATVS-FFp DB dataset for training and testing. Two different splits of the data have been tested in this workوThe first time, the proportions were training = 60%, valid = 20%, and testing= 20%, and in the second division, we reduced the images for verification, increased them in the test, and kept the training rate as it was, namely the valid = 10%, and testing = 30%. The model showed that it is affected by the number of images in the test and shows higher results the more the test phase is given. The trained model achieved 99% training accuracy and 97% testing accuracy.
Keywords: Fingerprint, Biometrics, Presentation Attack Detection (PAD), Presentation Attack (PA), Convolutional Neural Network (CNN).
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Black box algorithms have demonstrated their superiority in handling the sparse data that the modern Internet is saturated with. Transparency, which is essential for building trust in recommender systems, is missing, nevertheless. Collaborative Filtering techniques require lots of item s ratings for the model to function properly, moreover, item-side information greatly aids in creating interpretations because filtering models frequently rely on ratings. In our study, we suggest a new approach for producing justifications while maintaining high accuracy. The findings of the research demonstrate that our suggested solution performs better in terms of accuracy and transparency.
Keywords: Machine Learning, Collaborative Filtering, Recommender Systems, Explanations
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
The COVID-19 pandemic spread worldwide in the year 2020 and became a global health emergency. This pandemic has brought awareness that social distancing and quarantine are ideal ways to protect people in the community from infection. Therefore, Saudi Arabia used online learning instead of stopping it completely to continue the education process. This paper proposes to use machine-learning algorithms for Arabic sentiment analysis to find out what students and teaching staff thought about online learning during the COVID-19 outbreak. During the pandemic, a real-world data set was gathered that included about 100,000 Arabic tweets related to online learning. The overall goal is to use sentiment analysis of tweets to find patterns that help improve the quality of online learning. The data set that was collected has three classes: "Positive," "Negative," and "Neutral." Cross-validation is used to run the experiments ten times. Precision, recall, and F-measure was used to measure how well the algorithms worked. Classifiers, such as Support Vector Machines, K nearest neighbors, and Random Forest, were used to classify the dataset. Moreover, a detailed analysis and comparison of the results are made in this research. Finally, a visual examination of the data is made using the word cloud technique.
Keywords: sentiment analysis, machine learning, opinion mining, online learning
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
The artificial intelligence algorithm Generative Adversarial Networks (GAN) is excellent in creating works that simulate human output. As a result, many researchers have created impressive and satisfying art pieces such as images of non-existent people or expressive paintings. The Hijazi heritage is full of unique art forms, including the Rawashin that adorn Hijazi buildings. With the remarkable technical progress of recent years, it has become necessary to highlight this identity in a contemporary way. This work aims to exploit and explore the capabilities of artificial intelligence techniques and GAN networks in creating and producing innovative new shapes with regard to Rawashin (wooden windows). The aim is to integrate such shapes with Arabic lettering in order to produce unprecedented designs in terms of Hijazi buildings. This is done by training the machine using a dataset consisting of images of different building shapes containing Rawashin and some Arabic calligraphy using two types of GAN models. As a result, the model was able to learn and produce a new style of Rawashin.
Keywords: GAN, VQ-GAQN, Rawashin, Arabic calligraphy
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Saudi Arabia’s Vision 2030, developed in 2016, is not limited to diversifying the economy and reducing the dependence on the hydrocarbon sector. Artificial intelligence (AI) is among the main strategic priorities of Vision 2030. Saudi Arabia is planning to be the world leader in technology by putting AI at the heart of the country’s development and growth. This study looked at AI related research outputs produced by Saudi Arabia to calculate the growth rate of AI research over the years and to measure how much of this research is covered by altmetrics, which basically capture all kinds of research mentions in many online platforms, such as social media. We used the AI-SPedia knowledge base repository, which accommodates AI research around the world with all the details, including bibliometric and altmetric indicators. After running the appropriate SPARQL code, we retrieved about 4,433 AI publications that are published by Saudi organizations. This study showed that the growth rate of AI research produced by Saudi Arabia increased from 28.23% before 2016 to 47.07% over time. This big jump was attributed mainly to Saudi Vision 2030 and the funding by Saudi government sectors, such as the Ministry of Education. This growth will have a great impact on the quantity of AI research and the scientific community’s tangible work in the AI domain. Moreover, this paper shows that the AI research outputs from the 10 most productive Saudi universities have an average of around 31.74% of altmetric coverage. This low altmetric coverage indicates a need for suitable mechanisms to promote the dissemination of AI research through the most popular online platforms.
Keywords: Scholarly Publishing, Growth Rate, Bibliometrics, Altmetrics.
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
The digitization of textiles (textronics) has created new opportunities for integration with conformable sensors to enable unobtrusive, noninvasive, and continuous decoding of vital body signals. This article offers a thorough analysis of the Internet of Things' textile sensors and the fabrication techniques utilized for textronic sensors according to the form factors of fiber, yarn, fabric, and apparel. Based on their effects on the mechanical and electrical performance of the textronic sensor, it also highlights the necessity of pretreatment and conditioning reporting of the textile form-factors. The study comes to the conclusion that textile-based sensors in the form of wearable computing devices that can be attached to or worn on the human body can be utilized as smart sensing devices to access the mobile internet in addition to being able to send information. The next generation of human-computer interfaces may be built on top of these sensors. One of the factors influencing the development of textile sensors is the ongoing discovery of novel conductive materials.
Keywords: Internet of things, textile sensors, wearable sensors, smart textiles
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Abstract—In today’s world, the field of information technology is rapidly evolving. Maintaining security and privacy is a major problem for cyber professionals. According to studies, the quantity of new malware is rapidly increasing. A keylogger is a highly sophisticated malware that records every keystroke made on the machine, allowing the attacker the potential to steal enormous amounts of critically sensitive information invisibly without the authorization of the message’s owner. Identifying keylogger is important to avoid data loss and sensitive information leaking. Anti-viruses can detect keylogger via heuristic and behavior analysis, but if the keylogger is not a Known threat, antivirus or anti-malware software cannot detect it as a virus. Machine learning is effective in detecting malware. This paper seeks to detect each application’s set of rights and storage levels and distinguish between programs with proper access and keylogger applications that can misuse permissions. This keylogger detection technique is fully black-box; it is based on behavioral traits that are universal to all keyloggers and do not rely on the keylogger’s internal structure. In this research, a keylogger detection model has been proposed using machine learning to detect the keylogger and spyware. The model has been trained on keylogger and spyware data-set to identify the host behavior during keylogger running on the system. The results will be evaluated by using several metrics and presented based on the classification report and confusion matrix to identify system success in detecting keylogger spyware.
Keywords: Keylogger, Detection, Spyware, Machine Learning
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Education is a significant tool for change. It contributes to social stability, develops livelihoods, improves health, and motivates long-term economic growth. One of the critical aspects of the educational process is student assessment, which enables teachers to address current issues in education and measure teaching effectiveness and student performance. Correcting assessments takes teachers time and effort. This research proposed a solution to this issue by automatically correcting the true percent relative error equation with less human interference as well as decrease grading bias by using a machine learning approach. In order to use machine learning algorithms, a true percent relative error dataset were created. A total of 550 solutions by different numerical analysis students at Umm Al-Qura University were collected and labeled in this work. Then this research tried and tested six machine learning classification algorithms. The results showed that the gradient boosting classifier algorithm achieved the highest percentage in all metrics. It achied 86% in accuracy, 87% in precision, 86% in recall, and 86% in F1. While the K-neighbors obtained the lowest percentage in all metrics.
Keywords: Machine learning; Automatical Correction; Artificial Intelligence; Gradient Boosting Classifier; Numerical Analysis.
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Abstract—As moving to the EVs is essential due to the environmental cues, we need to solve the problems and undertake the challenges of the EVs to expedite this move. One of these challenges is the long recharging time compared to the traditional refueling process and the consequence of this issue is overcrowding in charging stations. To solve this problem, a charging management system is needed. However, an effective charging management system requires efficient communication networks that facilitate the exchange of information between the EVs, charging stations and the management server. In this work, we study the feasibility of exchanging EV’s data using V2V and V2I communications infrastructure considering AlKhobar city in Saudi Arabia as a case study. SUMO GUI and OMNET++ simulators have been used to set up the environment. The simulation results show that the minimum data throughput requirements are 0.83, 1.11, and 59.19 Bytes for V2V, V2I, and I2I respectively. This paper has contributed to finding the minimum requirement of the data throughputs of the communication links in vehicular communication.
Keywords: Data Exchange, RATs, Electric Vehicles, IoEV, Vehicle-to-Vehicle (V2V);
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
Currently, the need for real-time COVID-19 de- tection methods with minimal tools and cost is an important challenge. The available methods are still difficult to apply, slow, costly, and their accuracy is low. In this work, a novel machine learning-based framework to predict COVID-19 is proposed, which is based on rapid inpatient clinical tests of lung and heart function. Compared with current cognition therapy techniques, the proposed framework can significantly improve the accuracy and time performance of COVID-19 diagnosis without any lab or equipment requirements. In this work, five parameters of clinical testing were adopted; Respiration rate, Heart rate, systolic blood pressure, diastolic blood pressure, and mean arterial blood pressure. After obtaining results for these tests, a pre-trained intelligent model based on Random Forest Tree (RFT) machine learning algorithm is used for detection. This model was trained by about 13,558 records of the COVID- 19 testing dataset collected from King Faisal Specialist Hospital and Research Centre (KFSH&RC) in Saudi Arabia. Experiments have shown that the proposed framework performs highly in detecting COVID infections by 96.9%. Its results can be output in minutes, which supports clinical staff in screening COVID-19 patients from their inpatient clinical data.
Keywords: COVID-19, machine learning, Random Forest Tree
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
The future of smart homes is exciting. The number of smart devices connected to the Internet is supposed to increase from 31 billion in 2020 to 75.4 billion by 2025. These devices are increasingly vulnerable to cyberattacks because of the inherent connectivity in Smart Home. The attacker can use many techniques to compromise the system or cause damage to it such as ransomware, data and identity theft, DDoS attacks, etc. The preliminary results of the study show that people do not have the necessary level of culture to deal with attacks, also they are not aware of the potential security risks in their Smart Homes. They are not aware of the need for more tools to secure them. In light of this issue, this study propose a novel intrusion detection system for Smart Home environments. The proposed intrusion detection system will be based on two deep learning algorithms CNN and LSTM. To train and test the proposed model this study use a new dataset called TON-IoT specific to IoT environment and contains many records about many recent attacks in this particular network. The main goal of this study is to help the user monitors his Smart Home devices by detecting intrusions using a Deep Learning approach. Then, the proposed system will show to the user the detected intrusions through a dashboard. Upon the detection of an intrusion in a device, a notification will be shown to the user through the dashboard. Besides, the system will recommend some countermeasures to the user to help him harden his Smart Home and reduce some potential risks. The obtained results shows that the study model outperforms many existing models based on Machine learning algorithms.
Keywords: Smart-Home, Deep-Learning , IDS, Vulnerability,CNN algorithm.
|
View Poster |
|
Join the Presenter for a Live Discussion |
|
Email the Presenter |
