000			04296ntm a2200337 i 4500
999			_c102253 _d102259
003			MY-KuUP
005			20251125111025.0
006			t||||fr|||| 000 0
007			ta
008			250203t20242024my a|||fr|||| 000 0 eng d
020			_aTHE0010000 (Local) _qHardback
040			_aUMPSA _beng _cUMPSA _erda
090			_aFTKEE .M37 2024 r Thesis
100	0		_aMarzia Ahmed, _eauthor.
245	1	0	_aNew bio-inspired barnacle optimizers based least-square support vector machine for time-series prediction of pandemic outbreaks / _cMarzia Ahmed
264		1	_aKuantan, Pahang : _bUMPSA, _c2024
264		4	_c© 2024
300			_axiii, 132 pages ; _billustration ; _c30 cm. + _e1 CD-ROM
336			_2rdacontent _atext
337			_2rdamedia _aunmediated
338			_2rdacarrier _avolume
347			_2rda _atext file _bPDF
500			_aFaculty of Electrical & Electronics Engineering Technology
502			_aThesis (Doctor of Philosophy) -- Universiti Malaysia Pahang – 2024
504			_aIncludes bibliographical references
520	3		_aPandemic outbreaks like Coronavirus disease (COVID-19) present unprecedented challenges, demanding accurate time-series prediction models to understand disease dynamics and inform public health interventions. Traditional single machine learning models struggle to capture complex temporal patterns, especially considering the influence of vaccination campaigns on confirmed cases, leading to suboptimal predictions. The objectives of this research are threefold: firstly, the aim is to create new hybrid optimized machine learning models that improve prediction accuracy; secondly, to overcome limitations in the original Barnacle Mating Optimizer (BMO) and its variants, ensuring the development of robust prediction models; and thirdly, to incorporate vaccination data into the prediction models, enabling adaptation to the dynamic changes in pandemic scenarios. The methodology involves three main steps. Firstly, this thesis proposes hybrid optimized machine learning models by combining improved BMO variants with Least Squares Support Vector Machines (LSSVM) to effectively capture intricate temporal patterns. The BMO-Gauss-LSSVM, BMO-LevyLSSVM, Selective Opposition-Based BMO variants harness BMO's exploration and exploitation capabilities to optimize the hyperparameters of LSSVM, resulting in enhanced prediction performance. Secondly, this thesis invents the new bio-inspired Gooseneck Barnacle Optimization Algorithm (GBO), drawing inspiration from gooseneck barnacle mating behaviors, to address the limitations of the original BMO and its variants, aiming for more realistic and robust optimization. Lastly, this study model various factors influencing the transmission dynamics of COVID-19 within the BMO and GBO variants, refining their predictive ability. The study demonstrates GBO's superiority over previous algorithms, including BMO and its variants, and other established methods. Through real-world case studies and mathematical tests, GBO consistently outperforms alternative algorithms. In the BMO family, SO-BMO-LSSVM achieves the highest accuracy at 99.69%, while GBO-LSSVM surpasses this, reaching an impressive 99.78%. Other algorithms like SSA-LSSVM achieve 97.76%, MVOLSSVM records 97.426%, PSO-LSSVM achieves 99.53%, and ANN shows an accuracy of 89%. These findings underscore the efficacy of GBO and its competitive edge in achieving accurate time-series predictions when compared to a diverse set of algorithms. The groundbreaking outcome of this research lies in the transformative impact of enhanced BMO variants, along with the innovative new bio-inspired GBO algorithm and GBO-LSSVM models. These advancements revolutionize epidemiological forecasting and public health preparedness by delivering precise and adaptable time-series predictions for pandemic outbreaks. By integrating evolutionary optimization with advanced machine learning techniques, these models offer potent solutions to effectively tackle the challenges posed by pandemics, such as the complex dynamics of COVID-19.
610	2	0	_aFaculty of Electrical & Electronics Engineering Technology _xDissertations
650		0	_aUniversities and colleges _xDissertations
650		0	_aTheses
942			_2lcc _cTHESIS