Iraqi Journal for Computers and Informatics
Not a member yet
273 research outputs found
Sort by
Hybrid Machine Learning Approaches for 5G Traffic Prediction
Full text linkABSTRACT Accurate traffic prediction poses great difficulties because of the continuously increasing scale and diversity of 5G network traffic, which is driven by user demands. Moreover, certain characteristics of 5G traffic are constantly changing; thus, simulations using traditional models often lead to incorrect estimations or inefficient utilization of available resources. Consequently, we propose a hybrid machine learning model that integrates support vector machine (SVM) and decision tree algorithms to enhance efficiency of 5G traffic prediction. The structure of the hybrid model dynamically adjusts by adding or removing hidden layers and units within the network to improve prediction performance. The efficacy of the proposed model is evaluated using metrics like mean squared error, mean absolute error, and root mean squared error (RMSE). Findings show that the hybrid model consistently achieves lower error rates than SVM alone. Further performance enhancement of the hybrid model in predicting 5G traffic is also supported by comparisons of R-squared values against signal-to-noise ratios. These outcomes show the potential of the proposed method to improve traffic prediction accuracy in 5G networks, serving as a powerful tool for network control
Sentiment Analysis in Arabic Text and Emoji Using Deep Learning Methods
Full text linkلقد أدى ظهور وسائل التواصل الاجتماعي إلى تبسيط النشر السريع للتوضيحات حول الإعلان الشامل والأفلام والسياسة والاقتصاد. وقد أدى هذا النمو إلى زيادة في اتساع نطاق الموضوعات التي تمت تغطيتها. تم دمج تحليل المشاعر هذا، والذي يشمل العديد من الجوانب. تم دمج بيانات المسح الضخمة للغة العربية و OMCD مع بيانات من Twitter لإبلاغ هذه الدراسة. تم تنفيذ طرق تضمين الكلمات المختلفة، مثل Spacy (W2V) و FastText و Arabic Bidirectional Encoder Representation (AraBERT). في سياق نماذج تحليل المشاعر، تم استخدام الشبكة العصبية التلافيفية (CNN) والذاكرة الطويلة والقصيرة المدى (LSTM) والشبكة العصبية المتكررة (RNN). استند تقييم أداء النموذج إلى الدقة. أسفرت طرق التعلم العميق (DL) باستخدام مجموعة بيانات AST (المشاعر العربية على تويتر) عن معدلات دقة للنموذج بنسبة 72٪ و 95٪. وفي الوقت نفسه، وقعت معدلات الدقة لمجموعة بيانات OMCD (مجموعة بيانات التعليقات المغربية المسيئة) في نطاق 54٪ إلى 84٪.The advent of social media has simplified the rapid publishing of explanations on inclusive announcements, movies, politics, and the economy. This growth has led to an increase in the breadth of topics covered. This emotion analysis includes many aspects. Arabic and OMCD survey big data were integrated with data from Twitter to inform this study. Different word embedding methods were implemented, such as Spacy (W2V), FastText, and Arabic Bidirectional Encoder Representation (AraBERT). In the context of sentiment analysis models, convolutional neural network (CNN), long short-term memory (LSTM), and recurrent neural network (RNN) were employed. The evaluation of model performance was based on accuracy. Deep learning (DL) methods using the AST (Arabic sentiment Twitter) dataset yielded 72% and 95% model accuracy rates. The accuracy rates for the OMCD (Offensive Moroccan Comments Dataset) is a dataset containing offensive comments in the Moroccan dialect. dataset fell within the range of 54% to 84%
INTELLIGENT SURVEILLANCE SYSTEM FOR FIRE DETECTION USING YOLOV8
Full text linkThis study describes a lightweight deep learning model trained on a self-made image dataset taken inside farms and open areas of the Holy Shrine of Al-Hussainiya in the City of Karbala, Iraq. This dataset includes fire and smoke images taken using a Samsung A52S camera in different weather conditions. The overall goal is to create a fire detection system model that can successfully replace the existing physical sensor-based fire detectors and lessen the issues that come with such fire detectors, including false and delayed triggering. Another goal is to control fires on farms or open areas and prevent crop damage as much as possible. Previous studies were reviewed. Moreover, the architecture of the You Only Look Once version 8 (YOLOv8) model was briefly explained, and the results it achieved were compared with those achieved by previous versions. Then, the proposed system was trained and evaluated with the YOLOv8 large model. Results showed that the proposed system outperformed the rest of the current systems in mAP, which reached 98.5%
Review of Authentication Systems based on Electroencephalogram
Full text linkTraditional authentication methods, such as the use of passwords and fingerprints, are susceptible to the risks of theft, loss, and forgery. However, an innovative and secure alternative exists in the form of electroencephalogram (EEG)-based authentication systems, which operate by measuring distinctive brainwave patterns. This particular review undertakes a comprehensive analysis of the current state of EEG-based authentication, delving into its advantages, challenges, and potential future directions. In doing so, we examine the underlying principles governing the acquisition and processing of EEG signals, explore the various techniques employed for feature extraction and classification, and evaluate the performance of existing systems. Moreover, we emphasize the significant advantages offered by EEG-based authentication, including its exceptional accuracy, capacity for liveness detection, and robust resistance to spoofing attempts. Nevertheless, we must also acknowledge and address the various obstacles that must be overcome to facilitate wider adoption of this authentication method, encompassing concerns relating to hardware affordability, user acceptance, and data privacy. Finally, we outline a series of promising research avenues that can potentially address these challenges and unlock the complete potential of EEG-based authentication, thereby enabling secure and convenient access control across a diverse range of domains
Digital Image Forgery Detection And Localization Using The Innovated U-Net
Full text linkA reliable image copy–move forgery detection approach adaptable to different scenarios of tampering with color images is crucial for many applications. Different methods and solutions have been effectively proposed, but they are still subject to false positive/negative detections and cannot handle the variety of copy–move forgeries. In this paper, a machine learning model that combines ResNet 50 and U-net architectures for automatic image forgery detection in color image(s) is presented. The proposed system is inspired by the ResNet 50 architecture as an encoder and the U-Net architecture as a decoder. The encoder function implies applying convolution and normalizing for feature extraction. Conversely, the decoder functions is locating the spatial features. The decoder in the U-Net network comprises multiple decoder blocks, which are connected to corresponding encoder blocks by employing concatenate layers. A binary mask is then produced to represent the tampered regions in the image. Quantitative experimental results on two standard public datasets and a comparison with state-of-the-art methods demonstrate the effectiveness and robustness of the proposed model
Artificial Intelligence Systems and Medical Negligence: An Overview and Perspective of a Case Study in Ghana Civil Procedure Rules, 2004 (C.I. 47)
Full text linkObjective: This article discusses the evidentiary requirements for demonstrating scientific negligence under Ghana’s Civil Procedure Rules 2004 (C.I. 47) in the context of emerging artificial intelligence (AI) diagnostic and treatment structures.Method: Legal analysis examines gaps in satisfying burden of proof and standards of evidence, obstacles that restrict evidence collection on AI device deficiencies, and suggestions for adapting legal responsibility policies to AI’s technical opacity.Findings: The present inability to interrogate algorithms, limited access to proprietary training data and methods, lack of diagnosed standards of care for software-based decision-makers, and shortage of qualified professional witnesses pose massive evidentiary challenges for plaintiffs seeking to confirm AI negligence.Conclusions/Recommendations: Standards strengthening algorithmic transparency, auditability, and explainability could ease evidentiary burdens for affected patients. Strict liability schemes and IP protections balancing public safety and innovation aims need to be considered moving forward.Scientific Contributions: This work adapts traditional medical liability systems to today’s realities of increasing reliance on AI in health care and proposes several improvements
A Review of House Detection from High Resolution Satellite Images
Full text linkDetecting houses through high-resolution satellite images has received increasing attention in recent years and is considered a basic but difficult task in the field of remote sensing. Although, there are many methods that have been developed in this field, there is still a need for an in-depth review of recent articles about Extracting buildings and houses from high-resolution satellite images. This study aims to provide a comprehensive review of articles published in the scientific literature including developments that have occurred in recent years in this field. The topic of house detection is a popular, widespread and rapidly emerging research topic in domain of remote sensing. Because its importance in many fields, including drawing and updating urban maps, monitoring change, detecting damage resulting from environmental disasters, land use analysis, population estimation, and other applications. In addition to Availability of high-resolution images produced by the new generation to satellites. Topics of automatic detection of houses, object-based approaches, machine learning, and deep learning through high-resolution RGB images
Boosting Learning Algorithms for Chronic Diseases Prediction: A Review
Full text linkالخوارزمية المعززة هي مجموعة من تقنيات التعلم الآلي التي تعتمد على فكرة أن اكتساب المتعلم الضعيف لمصنفات أساسية متعددة قد يؤدي إلى نتائج تتفوق على نتائج أي مصنف بسيط يستخدم بمفرده. لا يوجد تقييم شامل لتقنيات التعزيز المستخدمة بانتظام ضد الأمراض التي تم التحقيق فيها بشكل كبير، على الرغم من حقيقة أن أساليب التعزيز قد استخدمت للتنبؤ بالمرض في العديد من الدراسات. وبالتالي فإن الغرض من هذا العمل هو إيجاد أنماط هامة في دقة أداء استراتيجيات التعزيز. ستساعد نتائج هذا العمل الأكاديميين على تحديد نهج معزز أكثر ملاءمة للتنبؤ بالمرض، بالإضافة إلى فهم أفضل للأنماط الحالية والنقاط الساخنة في نماذج التنبؤ بالأمراض التي تستخدم تعزيز التعلم. أظهرت النتائج أن خوارزمية adaboost تفوقت على الخوارزميات الأخرى من حيث الدقة، حيث حققت أكثر من 90%. توضح هذه المراجعة أيضًا كيف يمكن للجمع بين طريقتين للتعزيز أن يزيد من دقة المصنف الأساسي. وباستخدام AdaBoost وLightGBM، بلغت الدقة 99.75%. تم استخدام تقنيات XGBoost وGradient Boosting في الأبحاث بشكل متكرر أكثر من خوارزميات التعزيز الأخرى.Boosting algorithms are a set of machine learning techniques that are predicated on the notion that a weak learner\u27s acquisition of multiple basic classifiers might yield results that are superior to those of any one simple classifier used alone. A comprehensive evaluation of regularly used boosting techniques against highly investigated diseases is lacking, despite the fact that boosting approaches have been used for disease prediction in many studies. Thus, the purpose of this work is to highlight the main algorithms and strategies in the boosting learning. The results of this work will help academics identify a more appropriate boosting approach to predict disease, as well as better understand current patterns and hotspots in diseases prediction models that use boosting learning. The results showed that adaboost algorithm outperformed other algorithms in terms of accuracy, achieving above 90%. This review also demonstrates how combining two boosting methods can increase the basic classifier\u27s accuracy. By using AdaBoost and LightGBM, the accuracy reached 99.75%. XGBoost and Gradient Boosting techniques were employed more frequently in researches than other boosting algorithms
Develop Approach to Predicate Software Reliability Growth Model Parameters Based on Machine learning
Full text linkOne of the most important aspects in determining the quality of a software product before placing it on the market is its reliability. The main problem in creating effective software that satisfies the user preferences is that it must be highly reliable. One important factor that has a remarkable influence on the overall reliability of a system is its software. Reliability is a critical aspect of software quality, and the software industry faces many challenges in its quest to produce reliable software at scale. Reliability models are a basic method for quantitatively calculating software reliability. Thus, this paper inspects the reliability of software applications as a substantial feature of this application and helps determine the extent of software reliability in performing specialized functions. This goal is accomplished by calculating the parameters of software reliability growth models (SRGMs). The parameters are evaluated using three algorithms: machine learning decision tree (DT), support vector machine (SVM), and K-nearest neighbors (K-NN). Results show that the SVM model achieves the best mean square error
Optimizing Diabetic Retinopathy Classification with Transfer Learning: A Lightweight Approach Using Model Clustering
Full text linkAccurate and rapid classification of diabetic retinopathy is critically significant in order to prevent vision loss. The advent of artificial intelligence introduces novel and potent methodologies for enhancing the classification of diabetic retinopathy as derived from medical imaging. Due to the large size of the model make it unsuitable in real world. This research paper is dedicated to the classification of diabetic retinopathy utilizing constrained resources while achieving elevated accuracy levels. We implemented weighted clustering technique within deep convolutional neural networks and transfer learning architectures: VGG 19, DenseNet 121, and EfficientNet B6. To mitigate the challenge posed by considerable model sizes without sacrificing accuracy, the best fit results were observed with EfficientNet-B6, where applying weighted clustering reduced the model size by a factor of 12 while maintaining high accuracy results of 92% for the APTOS-2019 data. This underscores the efficacy of employing lightweight techniques to enhance the practicality of extensive models for the early diagnosis of diabetic retinopathy