Pusat Pengajian Sains Komputer - Tesis
Browse
Recent Submissions
- PublicationChildhood Obesity And Overweight Predictions Using Hybrid Data Mining Techniques(2013-02)Adnan, Muhamad Hariz MuhamadChildhood obesity has become a worrying global epidemic. Evidences show that childhood obesity persists into adulthood. Therefore, predicting obesity at an early age is both useful and important. The objectives of this study are to identify significant risk factors and protective factors of childhood obesity in Malaysia; to select and propose parameters for prediction; to investigate suitable data mining techniques for childhood obesity and overweight predictions; and to propose hybrid data mining techniques to increase the sensitivity of predictions. This study consisted of five stages that are risk factors reviews, data collections, parameter selection, empirical study, propose and evaluation of hybrid techniques. The factors that are related to the Malaysian children are: sex, catch-up growth, premature birth, adiposity rebound, breastfeeding duration, soup, sandwich, snack or chocolate eaten outdoors, excessive television (TV) watching, eating junk food, eat junk food in front of TV, warm meals for supper, eating fried food and fruits, physical activity, duration of sleep, number of sibling, and parents BMI. Three hybrid techniques have been proposed. The first technique is a hybrid of Naïve Bayes (NB) and Classification and Regression Tree (CART) in a NB-Tree framework; the second technique is using the CART for variable selection and NB for prediction (CART Selection-NB Classification); and the third technique integrates mean value calculation and Euclidean Distances classification into the second technique (Mean-Euclidean
- PublicationEnhanced Gray Wolf Algorithm Based Feature Selection For Sentiment Analysis(2025-08)Alqablan, Tamara Amjad AbdelkarimPeople access social media platforms through a range of IoT devices, including smartphones, tablets, and wearable devices, which lead to the generation of a significant quantity of data. The high dimensional of data is considered a major challenge in analyzing and extracting knowledge from big data. Feature selection is a technique concerned with selecting only the most relevant features to improve performance. In the context of wrapper-based feature selection approaches, the algorithm is often stuck in a poor solution (feature subset) due to its inability to efficiently explore the search space. To tackle this problem, first, an enhanced wrapper-based approach that combines the BGWO algorithm with the AβHC local search algorithm (Aβ-BGWO) was proposed to enhance the BGWO exploitation capabilities and improve its performance in terms of local search ability. Second, enhance the Aβ-BGWO by integrating it with BHHO (Aβ-BGWHHO) to improve the global convergence of A𝛽-BGWO, enhance its exploration capabilities, and fine-tune between exploitation and exploration. Third, due to the nature of the sentiment analysis datasets, filter methods can effectively identify and eliminate noise features, thereby enhancing the quality of feature subsets evaluated by the wrapper approach.
- PublicationLoad Balancing In A Distributed Environment : Implementation On Ethernet And Atm(1996-11)Chan, Huah YongLoad balancing is a technique for distributing processes amongst processors within a parallel computer or amongst workstations connected by a network in an effort to equalize workload. Job placement and migration are used to dynamically schedule processes. The objective is to maximize system throughput by optimizing the utilization of resources. ATM (asynchronous transfer mode) is of very high interest in the parallel computing community mainly because of its high and more reliable throughput. Various research results indicate that atm shows better performance when used with parallel computing and distributed processing[LIN94, HUAN94]. MPI (message passing interface) is a standard, portable and formally specified parallel (multicomputers) programming language. It uses a message passing model with an explicit control for parallelism
- PublicationApplication Of Genetic Algorithms In Optimizing The Distribution Of Examinations In An Examination Scheduling Problem(1996-09)Kilani, Yousef Moh' D AhmadThe scheduling of exams in universities and institutions is known to be a highly constrained problem. Soft and hard constraints need to be satisfied. The increasing number of students and courses offered from year to year increases the difficulty of scheduling the exams. Different approaches have been used to tackle the exam timetabling problem. These approaches include: heuristic approaches, mathematical approaches and random search approaches. Among the random search approaches is the genetic algorithms. This project investigates the use of genetic algorithms on the problem of scheduling exams for the universiti sains malaysia. The algorithm described in this thesis improves the checking of consecutive exams. Moreover, the thesis investigates the performance of many genetic algorithms and it looks at different possibilities to show the behavior and performance of the genetic algorithm. Comparison is done with the current method and a hybrid approach is shown to generate better result.
- PublicationConditional And Anonymous Payment Protocol With Local Dispute Resolution Mechanism In Payment Channel Networks(2025-08)Hasan, Hasan Hasan FalahThis research proposes a conditional and anonymous payment protocol with a local dispute resolution mechanism, addressing three key issues. First, a lightweight and Lightning Network-compatible Onion Style Symmetric Commitment (OSSC) that ensures a secure and efficient locking and unlocking scheme. Second, the Efficient Multi-party Conditional Payment Flow (EMCPF) protocol relies on a Hybrid Encryption Mechanism (HEM). It extends a refined OSSC version by avoiding the ideal anonymous communication assumption, preserving privacy, security, and preventing long collateral durations and Wormhole attacks. Finally, a Local Dispute Mechanism (LDM) is extended with EMCPF to resolve disputes and unresponsiveness cases between channel parties locally and synchronously, avoiding costly blockchain lifecycle.