Research Title: Sentiment Analysis of COV19 Impact on Education and Economy Sectors using Arabic Tweets

Author: Aya Adnan Oqla Miqdady, Published Year: 2023

2023 14th International Conference on Information and Communication Systems (ICICS),

Faculty: Information Technology

Abstract: The coronavirus (COVID-19) infected millions of people around the world. Due to the disease's rapid spread, all governments have decided to implement lockdowns. The two key pillars of any nation, education and economy were negatively impacted. This study investigates public opinion on COVID-19 effects on these two key sector by collecting Arabic tweets from Twitter from the MENA region. Sentiment analysis was applied using the collected dataset for each sector separately. We created five cutting-edge deep learning models to apply them in the prediction process. Such models named Convolutional Neural Network (CNN), Long Short Term Memory (LSTM), LSTM plus CNN, BERT, and AraBERT. The outcomes showed that in two datasets, the LSTM with CNN and LSTM provided the greatest performance results in the education data set, while BERT and AraBERT has the best performance in the economy dataset. The results presented that these algorithms created information that helped several government decision-makers make important and wise decisions on the standard of living in the Arab World Country.

Keywords: COVID-19 , Deep learning , Economics , Sentiment analysis , Social networking (online) , Biological system modeling , Education

Research Title: Predicting Bug Severity Using Machine Learning and Ensemble Learning Techniques

Author: Aya Adnan Oqla Miqdady, Published Year: 2023

2023 14th International Conference on Information and Communication Systems (ICICS),

Faculty: Information Technology

Abstract: Software quality can be adversely affected by various bugs in the system. Bug identification and fixing are part of every software development life cycle. However, in practice, large and long-lived systems may encounter an enormous number of bugs during their lifetime. Additionally, identifying and fixing a large number of software bugs requires a vast amount of allocated resources that may hinder the software budget. Thus, proper management including bug classification and prioritization is needed. Therefore, this paper employs modern Machine Learning techniques to help developers identify and classify bugs based on various factors including their severity level. Six different datasets are used to evaluate the proposed ML models. Features are extracted from bug descriptions using natural language processing techniques like TF-IDF. Evaluated models include K-Nearest Neighbors, Naïve Bayes, Neural Networks, Random Forest, Support Vector Machine, and various ensemble learning methods. All candidate models are evaluated based on their accuracy for each of the used datasets. Results show that the Neural Networks model exceeds on all six datasets, achieving an accuracy that ranges from 93% to 95%. It also exceeds in other measures as well including precision, recall, and F1-Score. These results indicate that the proposed method performs satisfactorily when identifying bugs in source code.

Keywords: Support vector machines , Source coding , Computer bugs , Neural networks , Software quality , Feature extraction , Software systems

Research Title: A Deep Learning Approach for Varicocele Detection from Ultrasound Images

Author: Aya Adnan Oqla Miqdady, Published Year: 2023

2023 14th International Conference on Information and Communication Systems (ICICS),

Faculty: Information Technology

Abstract: Varicocele is a disease characterized by abnormal dilatation of the scrotal venous plexus pampiniformis. It is a common cause of infertility and may cause pain or discomfort in some cases. In this article, we present a new approach for automatic classification of varicocele using Deep Learning convolutional neural networks. The available dataset consists of images converted to different color modes. Each color mode dataset is partitioned, augmented, and trained using a Deep Learning network. Experiments were conducted with all possible combinations of four different pre-trained models and over three color modes to determine the best combination that achieves the highest performance, with and without augmentation. Training and testing were evaluated using various metrics. Analysis of the results demonstrated the effectiveness of the proposed system. Results showed an accuracy of 83.1% with the RGB color mode when using ResNet50, and that the effect of augmentation was small.

Keywords: Deep learning , Measurement , Training , Ultrasonic imaging , Image color analysis , Pain , Gray-scale

Research Title: Pronoun Replacement Approach for Enhancing Arabic Text Summarization

Author: Aya Adnan Oqla Miqdady, Published Year: 2022

2022 13th International Conference on Information and Communication Systems (ICICS),

Faculty: Information Technology

Abstract: The world witnessed massive and unprecedented growth in the amount of information available online. Identifying and consuming needed content has been difficult and time-consuming. Therefore, automatic text summarization to obtain accurate and relevant collections of the vast amount of available information is essential. Methodologies for automatic summarization of Arabic text remain immature, due to the complexity inherent in the Arabic language in terms of structure and morphology. The summarization process is based on the frequency which the trainer used to fetch important sentences to compute the keywords weightage that aims to construct the summary. This study proposes replacing pronouns with proper nouns before applying the Arabic text summarizer. This enhances the quality of the summarization process and helps reader better understand the summary.

Keywords: Natural Language Processing (NLP) , Text Summarization , Pronoun Replacement

Research Title: A CNN and Image-Based Approach for Malware Analysis

Author: Aya Adnan Oqla Miqdady, Published Year: 2022

2022 International Conference on Emerging Trends in Computing and Engineering Applications (ETCEA),

Faculty: Information Technology

Abstract: Malware attacks have various types, patterns, and volumes and have become more sophisticated and severe. Using machine learning to classify and detect malware is one of the approaches to mitigate malware attacks. However, malware classification suffers from some challenges such as the time required in manipulating a huge number of malware files. This paper proposes a Convolutional Neural Network (CNN) model and a pre-processing approach to solve the aforementioned issue. The contribution of this paper is based on converting the dataset into RGB images followed by scale maximization step. In addition, the paper proposes a preprocessing approach for the input datum of images. The results prove that the proposed preprocessing methods have a strong impact on enhancing the overall accuracy by increasing the accuracy from 92.5% to 98%.

Keywords: Convolution , Neural networks , Data preprocessing , Machine learning , Gray-scale , Market research , Malware

Research Title: Autokeras Approach: A Robust Automated Deep Learning Network for Diagnosis Disease Cases in Medical Images

Author: Aya Adnan Oqla Miqdady, Published Year: 2023

Journal of Imaging,

Faculty: Information Technology

Abstract: Automated deep learning is promising in artificial intelligence (AI). However, a few applications of automated deep learning networks have been made in the clinical medical fields. Therefore, we studied the application of an open-source automated deep learning framework, Autokeras, for detecting smear blood images infected with malaria parasites. Autokeras is able to identify the optimal neural network to perform the classification task. Hence, the robustness of the adopted model is due to it not needing any prior knowledge from deep learning. In contrast, the traditional deep neural network methods still require more construction to identify the best convolutional neural network (CNN). The dataset used in this study consisted of 27,558 blood smear images. A comparative process proved the superiority of our proposed approach over other traditional neural networks. The evaluation results of our proposed model achieved high efficiency with impressive accuracy, reaching 95.6% when compared with previous competitive models.

Keywords: Artificial intelligence (AI); convolutional neural network (CNN); deep learning (DL); malaria parasites

Research Title: Autokeras Approach: A Robust Automated Deep Learning Network for Diagnosis Disease Cases in Medical Images

Author: Aya Adnan Oqla Miqdady, Published Year: 2023

Journal of Imaging,

Faculty: Information Technology

Abstract: Automated deep learning is promising in artificial intelligence (AI). However, a few applications of automated deep learning networks have been made in the clinical medical fields. Therefore, we studied the application of an open-source automated deep learning framework, Autokeras, for detecting smear blood images infected with malaria parasites. Autokeras is able to identify the optimal neural network to perform the classification task. Hence, the robustness of the adopted model is due to it not needing any prior knowledge from deep learning. In contrast, the traditional deep neural network methods still require more construction to identify the best convolutional neural network (CNN). The dataset used in this study consisted of 27,558 blood smear images. A comparative process proved the superiority of our proposed approach over other traditional neural networks. The evaluation results of our proposed model achieved high efficiency with impressive accuracy, reaching 95.6% when compared with previous competitive models.

Keywords: Artificial intelligence (AI); convolutional neural network (CNN); deep learning (DL); malaria parasites

Research Title: DRDM: Deep Learning Model for Diabetic Retinopathy Detection

Author: Aya Adnan Oqla Miqdady, Published Year: 2024

International Conference in Emerging Technologies for Sustainability and Intelligent Systems (ICETSIS),

Faculty: Information Technology

Abstract: The application of Artificial Intelligence is being applied in the medical industry at a quick pace, and it is currently serving as the main source of support for clinical practice solutions. Clinical practice accuracy could be improved and costs could be decreased with the use of deep learning techniques. To diagnose Diabetic Retinopathy, an effective and dependable method for automatic screening must be identified. However, deep-learning models may face difficulties due to a lack of data in several medical fields. The Diabetic Retinopathy Detection Model (DRDM), a deep learning model, is proposed in this research to identify retinal images as either infected or uninfected. The data transformation approach is used to address the lack of Diabetic Retinopathy data, which helps prevent overfitting by doubling the data. The paper shows that building a highly complex model like EfficientNetB3 or VGG16 is not necessary to achieve high performance, where, the experiment's test results approved that the DRDM model outperforms such pre-trained models. Furthermore, it took much less time for the DRDM model to produce these results.

Keywords: Deep learning , Industries , Diabetic retinopathy , Developing countries , Retina , Data models , Sustainable development

Research Title: An Adaptive Query Approach for Extracting Medical Images for Disease Detection Applications

Author: Aya Adnan Oqla Miqdady, Published Year: 2024

Arabian Journal for Science and Engineering,

Faculty: Information Technology

Abstract: Different applications heavily benefit from automatic deep learning including image classification, segmentation, and analysis. It significantly adds value to clinical systems through computer-aided detection, curing planning, diagnosis, and therapy through the acquisition of the most informative images. However, this deep learning approach faces one of the main hurdles in image processing: the necessity of a large, labeled dataset. Actually, such requirements in medical image analysis applications are considered excessively costly to acquire. Active learning methods can mitigate such issues by reducing the number of annotated images while raising the model’s performance. This paper introduces an active learning framework based on a novel sampling technique, where it queries the unannotated samples that behave differently from current training set samples. The adaptive sampling method is optimized by stochastic gradient descent approximation. This optimization leads to the construction of an adaptable and robust system that meets the needs of medical control systems. Moreover, such novelties contribute to a respectful enhancement of the model’s deep network performance when training over a few numbers of annotated images to reach underlying accuracy. The proposed structure outperforms other AL methods, as proved by the experimental results using stochastic gradient descent optimization technique over Skin Cancer, Pediatric Pneumonia, and COVID-19 datasets, which achieved an accuracy of 72.5%, 90%, and 90.5 using only 42.8%, 8%, and 5% human-labeled training data, respectively.

Keywords: Deep Learning, Active Learning, Sampling Technique, Stochastic Gradient Descent, Medical Image

Research Title: ArSa-Tweets: A Novel Arabic Sarcasm Detection System Based on Deep Learning Model

Author: Aya Adnan Oqla Miqdady, Published Year: 2024

Heliyon,

Faculty: Information Technology

Abstract: Sarcasm in Sentiment Analysis (SA) is important due to the sense of sarcasm in sentences that differs from their literal meaning. Analysis of Arabic sarcasm still has many challenges like implicit indirect idioms to express the opinion, and lack of Arabic sarcasm corpus. In this paper, we proposed a new detecting model for sarcasm in Arabic tweets called the ArSa-Tweet model. It is based on implementing and developing Deep Learning (DL) models to classify tweets as sarcastic or not. The development of our proposed model consists of adding main improvements by applying robust preprocessing steps before feeding the data to the adapted DL models. The adapted DL models are LSTM, Multi-headed CNN-LSTM-GRU, BERT, AraBert-V01, and AraBert-V02. In addition, we proposed ArSa-data as a golden corpus that consists of Arabic tweets. A comparative process shows that our proposed ArSa-Tweet method has the most impact accuracy rate based on deploying the AraBert-V02 model, which obtains the best performance results in all accuracy metrics when compared with other methods.

Keywords: Deep learning (DL), Sarcasm, Sentiment analysis (SA), Machine learning, Natural language processing (NLP), Tweets