Safar Maghdid Asaad

The field of natural language processing (NLP) has undergone significant expansion over the last decade. Many human-being applications are conducted daily via NLP tasks, starting from machine translation, speech recognition, text generation and recommendations, Part-of-Speech tagging (POS), and Named-Entity Recognition. However, low-resourced languages, such as the Central Kurdish language (CKL), remain largely unexamined due to the richness of their language and the shortage of necessary resources to support their development. The POS tagging task is the foundation of other NLP tasks; for example, the POS tag set has been used to standardize languages, providing the relationship between words within sentences, which is then followed by machine translation and text recommendation. Specifically, for the CKL, most of the utilized or provided POS tagsets are neither standardized nor comprehensive. To this end, this study presents an accurate and comprehensive POS tagset for the CKL, aiming to improve the performance of Kurdish NLP tasks. The article also collected most of the POS tags from various studies, as well as from Kurdish linguistic experts, to standardize part-of-speech tags. The developed POS tagset is designed to annotate a large CKL corpus, supporting Kurdish NLP tasks. The initial investigations of this study, conducted by comparing the listed POS tagset with the Universal Dependencies framework for standard languages, indicate that it can streamline or correct sentences more accurately for Kurdish NLP tasks.

Cepstral features, such as Gammatone Cepstral Coefficients (GTCC), have recently been applied in fault detection and diagnosis. However, GTCC was originally designed for speech feature extraction rather than fault detection, which limits its ability to effectively capture relevant features for fault identification. In this research, three key parameters of GTCC namely, the number of coefficients, maximum frequency, and minimum frequency are optimized using two metaheuristic algorithms: Fitness Dependent Optimizer (FDO) and Grey Wolf Optimization (GWO). These parameters are vital for enhancing the effectiveness of GTCC in extracting relevant features from the vibration signal for gear defect identification. Specifically, the maximum and minimum frequency values are critical for capturing the Gear Mesh Frequency (GMF), a common indicator of gear faults. Furthermore, optimizing the number of GTCC coefficients helps reduce model complexity. Experimental results demonstrate that optimizing GTCC parameters with GWO improves fault detection performance using an SVM classifier, achieving over 1% and 3% accuracy improvements on the PHM09 and DDS datasets, respectively.

- The field of natural language processing (NLP) has dramatically expanded within the last decade. Many human-being applications are conducted daily via NLP tasks, starting from machine translation, speech recognition, text generation and recommendations, Part-of-Speech tagging (POS), and Named-Entity Recognition (NER). However, low-resourced languages, such as the Central-Kurdish language (CKL), mainly remain unexamined due to shortage of necessary resources to support their development. The POS tagging task is the base of other NLP tasks; for example, the POS tag set has been used to standardized languages to provide the relationship between words among the sentences, followed by machine translation and text recommendation. Specifically, for the CKL, most of the utilized or provided POS tagsets are neither standardized nor comprehensive. To this end, this study presented an accurate and comprehensive POS tagset for the CKL to provide better performance of the Kurdish NLP tasks. The article also collected most of the POS tags from different studies as well as from Kurdish linguistic experts to standardized part-of-speech tags. The proposed POS tagset is designed to annotate a large CKL corpus and support Kurdish NLP tasks. The initial investigations of this study via comparison with the Universal Dependencies framework for standard languages, show that the proposed POS tagset can streamline or correct sentences more accurately for Kurdish NLP tasks.

Recently, a massive amount of critical data and immense sensitivity stemming from the interconnectivity of billions of devices (such as wireless sensing, embedded technologies, and radio frequency identification) have leveraged the diverse capabilities of modern technologies such as the Internet of Things (IoT). Owing to their limited capabilities, it is crucial to prioritize protecting the sensitive data generated by these devices. Traditional encryption methods, such as AES and RSA, often take too long to run and require excessive memory for devices with limited resources to handle well. In addition, designing a simple encryption algorithm should not affect the integrity and security of the data. To this end, the proposed algorithm uses an 80-bit cipher key to encrypt a 64-bit data block. In addition, it improves the complexity of security by using Feistel and Substitution-Permutation (SP) architectures to achieve complex encryption in only six rounds. The proposed algorithm was simulated using MATLAB and FELICS tools. The algorithm was implemented on both image and text data. Simulations demonstrated that the proposed algorithm achieves superior complexity, security, and robustness against various attack types, making it highly suitable for IoT environments.

Machine learning (ML) approaches cover several aspects of daily life tasks, including knowledge representation, data analysis, regression, classification, recognition, clustering, planning, reasoning, text recommendation, and perception. The ML approaches enable applications to learn and adapt with or without being directly programmed from previous data or experience. The ML techniques, coupled with current technologies, provide a range of solutions, starts from vision-based applications to text-generation applications. To this end, this article presents a comprehensive overview of the approaches of ML, including supervised, unsupervised, semi-supervised, reinforcement, and self-learning. This review critically examines the roles performed by these aforementioned approaches in terms of their weaknesses and strengths. Furthermore, within this study, a new comparative analysis is conducted by reviewing existing studies and evaluating ML techniques using metrics including data requirement, accuracy, complexity, interpretability, scalability, applications, and challenges. Thereafter, the implemented ML techniques are classified, and their key findings are examined. The comprehensive review demonstrates that neither standalone nor hybrid ML techniques can completely satisfy all of the evaluated metrics, the necessity of customized solutions based on the requirements of particular applications.

Electrocardiogram (ECG) analysis is widely used by cardiologists and medical practitioners for monitoring cardiac health. A high-performance automatic ECG classification system is challenging because there is difficulty in detecting and categorizing different waveforms in the signal, especially in manual analysis of ECG signals, which means, a better classification system is needed in terms of performance and accuracy. Hence, in this paper, the authors propose an accurate ECG classification and monitoring system called convolutional neural network-k nearest neighbor (CNN-kNN). The proposed method utilizes 1D-CNN and kNN. Unlike the existing techniques, the examined technique does not need training during classifying the ECG signals. The CNN-kNN is evaluated against the PhysioNet’s MIT-BIH and PTB diagnostics datasets. The CNN is fed using the ECG beat raw signal directly. In addition, the learned features are extracted from the 1D-CNN model and its dimensions are reduced using two fully connected layers and then fed to the k-NN classifier. The CNN-kNN model achieved average accuracies of 98% and 97.4% on arrhythmia and myocardial infarction classifications, respectively. These results are evidence of the great ability of the proposed model compared to the mentioned models in this article.

The Internet-of-Things (IoT) has been used with greater frequency to track peoples’ daily activities, particularly those conducted indoors. Wi-Fi technology has been also been used as an alternative to global navigation satellite system (GNSS) technologies to track indoor activities. The received signal strength indicator (RSSI) is widely used to assist in the positioning of Wi-Fi signals. However, the RSSI-based technique suffers from multipath, non-line-of-sight (NLOS) problems and the fluctuation of RSSI measurements via Wi-Fi chipsets. One of the most well-known RSSI-based approaches is to apply the fingerprinting method to do the positioning. However, the fingerprinting-based form has an additional problem due to the lack of RSSI data samples, specifically in harsh area with a huge number of classes or reference points (RPs) and an unstable matching process algorithm. To mitigate the problems of the RSSI-based fingerprinting approach, this research proposes a novel matching process algorithm called Norm_MSATE_LSTM. We first performed the augmentation process to increase the RSSI data records via the Mean Stander deviation Augmentation TEchnique (MSATE). The RSSI records were normalized (norm), and the long short-term memory (LSTM) technique was applied to estimate the correct positions. Finally, the proposed matching algorithm was compared with the stand-alone matching algorithms, including the weighted k-nearest neighbors (WkNN) and LSTM. The results obtained from the experiments and the simulated experiments using OMNeT++ show that the proposed matching algorithm may improve positioning accuracy by 33.1% and 57.5% when only augmentation and augmentation with normalization are applied, respectively.

The demand for using location-based services (LBSs) is rapidly increased, specifically in the last decade. Most people’s daily activities are related to LBS services, including navigation, billing address, tracking stuff, transportation, and other point-of-interest (POI). In the same manner, many solutions are widely available to process the positioning from outdoors to indoors. One of the most utilized positioning solutions is using fingerprinting-based techniques via different technologies, including WiFi, Bluetooth, 3G/4G, and UWB. Many attempts have been made to enhance fingerprinting-based positioning and then to provide an accurate solution. The recent attempts are referred to use modern machine learning algorithms as fingerprinting matching process. However, there is no single solution to provide an accurate, low-cost, on-the-go, and seamless positioning solution. Therefore, this article aims to address the issues of using fingerprinting-based positioning. A new taxonomy for the recent solutions, which are related to fingerprinting-based techniques, is also designed. Accordingly, machine learning algorithms which have been used in fingerprinting-based technique and their challenges are investigated.