https://repository.sustech.edu/handle/123456789/1290 Thu, 09 Apr 2026 19:27:48 GMT 2026-04-09T19:27:48Z https://repository.sustech.edu/handle/123456789/28398 Automatic Recognition and Identification for Mixed Sudanese Arabic – English Languages Speech Elfahal, Mohammed Osman Eltayeb; Supervisor, Mohammed Elhafiz Mustafa; Co-Supervisor, Rashid A. Saeed Mixed speech is the phenomena of using more than one language in a single sentence, this occurs in communication between bilinguals to express their ideas and thoughts using vocabulary of both languages, even occurs among none bilingual people to describe product originally from second language. This thesis addresses the problem of mixed speech communication in multilingual communities. This is regional problem faces shortage resources and studies. Sudanese Arabic and English languages are the two languages selected for this research to build a generalized mixed speech and language identification model, the first is common and formal language among the Sudan and the latter is international, language of science and primary lesson in Sudan education systems. For experimental purposes, mixed speech corpus was built including most frequent daily life Sudanese Arabic and English mixed sentences, collected through social media applications campaign, 75% of this collection is read by 87 bilingual Arabic natives in office environment resulting in 2289 audio files associated with their transcription for training purpose, considering speakers and code-switch types, environment as factors affecting performance of the model at recording time. Based on the assumption that native language dominance others in mixed speech, proposed solution for generalizing recognition model is centered around Sudanese Arabic language. The solution keeps the original words for each language participates in switching in all components of the model such as mixed phonetic dictionary, mixed languages lexicon, etc., except for Acoustic Model (AM) Arabic language is used instead of its original language based on assumption that native speaker does not suddenly reconfigure his articulation organs to produce sounds as natives do. Open source CMU SHPINX is adapted for this mixed speech task, proposed model, which is consider effected by native language dominance, outperforms existing single pass and multi pass models achieving overall accuracy of 33.05% in term of Word Error Rate (WER). Mixed speech produce hybrid language not belong to each participating language, interface for further linguistic computation is provided to deal with this new language. The interface contains recognized word, its order in the sentence, recognition confidence and its language identity. Language identification in the model is simply looked up identity from mixed languages lexicon to avoid effects of unclear language discrimination attributes in such speech. Achieved results, prove the possibility to generalize the model based on Arabic language, module for phonemes clustering and comparison needed to serve as front-end to detect new language phonemes that are not included in phonemes set in order to add new language to the model. Sun, 25 Aug 2019 00:00:00 GMT https://repository.sustech.edu/handle/123456789/28398 2019-08-25T00:00:00Z https://repository.sustech.edu/handle/123456789/28311 A Safety-Based Architectural Design Method for Software Product Lines Suliman, Mozamil Ebnauf Elgodbe; Supervisor, - Hany H. Ammar; Co-Supervisor, - Aisha Hassan The safety is considered one of the most critical issues in the design of the modern systems (e.g. cyber-physical systems). With the increasing attention of software safety, how to improve software safety has already become a more important concerned issue, especially for the safety-critical systems. The Software Product-Line (SPL) and reusable software components are suitable approaches for these systems, which are often re-engineered from existing systems. Currently, the influence of the architecture in assurance of software safety is being increasingly recognized. However, the safety-based architectural design methods are limited in SPLs because of the complexity and variabilities existing in SPL architectures. For that, this work seeks to find an efficient and effective method that can be used into the design process of the safety-critical SPLAs which enhances and manages the safety of SPLs. The work proposed a method for safety-driven software product line architecture design (SSPLA). For efficiency, a number of efforts have been made. In this context the proposed design method mentioned above is configured and adapted to be state-based architecture design method. Also as a pattern based development of the reference architecture can support the development and application process of the product lines a new safety design pattern of statechart is developed. The result is an object-oriented design pattern which handles the safety attribute. Additionally, as there is a tight interplay between safety and security, and in order to address the influence of the security issues in the safety design using patterns, a pattern development approach is proposed which is then used to enhance the proposed safety design pattern of statechart. In order to show the applicability of our work as well as evaluate it, a simplified safety assessment model is developed as well as using of two case studies. The evaluation results show that there is a considerable improvement in the safety design of the SPLA after applying our work. The results have proved that the state-based approach highly supports the development of the safety critical systems and it is effective to handle the safety and security together in the design of the safety pattern which provides more benefits as it is a high level reuse. Finally, this research will benefit both architects and safety engineers who can design SPLAs or develop software products. Thesis Mon, 12 Dec 2022 00:00:00 GMT https://repository.sustech.edu/handle/123456789/28311 2022-12-12T00:00:00Z https://repository.sustech.edu/handle/123456789/28258 Enhanced Deep Learning Framework for Face Verification Across Age Progression Osman, Areeg Mohammed; Supervisor, -Serestina Viriri, Professor Facial aging is a texture and shape variations that affect the human face as time progresses. Current face verification across age systems lack the required efficiency to recognize facial shape and texture variations at the same time while maintaining high accuracy, so the need was to create a powerful model that could identify these variations efficiently. Currents frameworks focus on using handcrafted techniques only, while others focus on the use of pre-trained models, so there is a need to develop an efficient model to extract shape and texture features in addition to taking advantage of the characteristics and strengths of handcrafted systems and pre-trained systems accordingly. The main objective of this research is to develop a model capable of extracting both shape and texture variations from the facial image, by fusing both shape and texture descriptors with pre-trained deep learning model to obtain better accuracy. Sequentially, a new model was developed from scratch using deep learning capable of extracting the variations that occur on the face. The research explores the use of a deeper convolutional neural network model from scratch, with Histogram of Oriented Gradients (HOG) descriptor to handle feature extraction and classification of two face images with the age gap. We studied the effect of fused GoogLeNet pre-trained convolution network model with Histogram Orientation Gradient (HOG) and Local Binary Pattern (LBP) feature descriptors at decision level through Majority Voting technique to achieve a good performance of our proposed system for face verification The experiments are based on the facial images collected from MORPH and FG-NET benchmarked datasets. Combining deep CNN with LBP seems to give minimum accuracy than combining it with both LBP and HOG. On the other hand, combining deep CNN architecture with HOG proved to give the highest accuracy value, which is 99.85%. Despite the FG-NET dataset contains fewer images, it appears that there is no improvement in the accuracy of the MORPH dataset. The future work is to implements a deeper pre-trained convolutional neural network models to make a comparison, also conduct a fusion of these models at decision level to improve accuracy. Thesis Thu, 19 Jan 2023 00:00:00 GMT https://repository.sustech.edu/handle/123456789/28258 2023-01-19T00:00:00Z https://repository.sustech.edu/handle/123456789/28134 A Machine Learning Holistic Strategy for miRNA-mRNA Module Discovery Shommo, Ghada Ali Mohamed; Supervisor, -Bruno Apolloni microRNA (abbreviated miRNA) is a small non-coding molecule, that is made up of approximately 22 nucleotides, found in plants, animals, and some viruses. They act as regulators by binding their targets to degrade or suppress the translation of their transcripts. Therefore miRNAs plays an important role in gene regulatory networks, and an improved understanding of miRNAs will widen our knowledge of these networks and their relation with diseases. A single miRNA targets multiple mRNAs, and a single mRNA is targeted by multiple miRNAs to develop a many-to-many relations, known as miRNAm- RNA module. Some methods have ignored this relation, by just focusing on miRNA-mRNA pairs. However current methods evolved to consider this relation but unfortunately they focused on some part of the data analyzed, and ignored the other. Unfortunately, they still leave open issues, but the higher benefit is that they provided results, that opened issue of the possibility of widening the scope of module discovery, so as to be extended to a wider disease spectrum, where miRNA-mRNA interactions play a relevant role. The research proposes a holistic procedure for miRNA-mRNA module identification that exploits as much data as possible. It uses machine learning and mathematical approaches to aid in the analysis and implementation. We adopt the strategy of postponing any decision until biological results are exploited. Many statistical tests have been diverted into specially-devised evolving metrics, for sake of possible solutions. Consequently, the Implementation on High Performance Computing (HPC) is crucial, since this strategy is rather expensive in terms of computation. Fortunately, it allows the discovery of modules whose miRNAs and mRNAs are not differentially expressed and the discovery miRNA targets not yet considered, as well. In this research, the procedure, is implemented on a Multiple Myeloma dataset publicly available on Gene Expression Omnibus (GEO) platform, as a case study of diseases, specifically as a cancer instance analysis, and scout some biological issues. The procedure has introduced novel strategies for miRNA-mRNA module discovery. Main achievements of this thesis work are: 1)we introduce a novel strategy for miRNA-mRNA module discovery; 2) we establish an unprecedented way of jointly using using many metrics to find new links between miRNA and mRNA clusters involving non differentially expressed RNA pairs as well; 3) and finally, we highlight new miRNA-mRNA interactions with a methodology that can be extended to a wide spectrum of diseases. Thesis Thu, 22 Dec 2022 00:00:00 GMT https://repository.sustech.edu/handle/123456789/28134 2022-12-22T00:00:00Z