FOLORUNSO, FAVOUR BOLADE and Covenant University, Theses (2023) PERFORMANCE EVALUATION OF TRANSFER LEARNING ALGORITHMS WITH OPTIMISERS FOR CLASSIFYING MALARIA CELL IMAGES. Masters thesis, COVENANT UNIVERSITY.
![[img]](http://eprints.covenantuniversity.edu.ng/style/images/fileicons/application_pdf.png) |
PDF
Download (280kB) |
Abstract
Malaria, a devastating disease transmitted by mosquitoes, continues to impose a significant global health and economic burden. Despite this, traditional diagnostic methods such as microscopy encounter limitations due to human error and high workload. Accurate identification of malaria-infected red blood cells is pivotal for effective management, and machine learning—particularly transfer learning algorithms—has shown promise in enhancing diagnosis. However, the optimal pairing of transfer learning architectures with suitable optimizers for precise classification remains unclear. To tackle these challenges, this study presents a performance evaluation of transfer learning algorithms alongside optimizers for classifying malaria red blood cell images. The study utilizes five transfer learning algorithms, including DenseNet201, ResNet50, VGG19, VGG16, and Xception, to investigate their performance in malaria red blood cell image classification. The dataset comprises 27,558 cell images, divided into Parasitized and Uninfected categories with 13,779 images each, sourced from the National Institutes of Health (NIH) dataset. Data pre-processing involves resizing, data splitting, label encoding, and data augmentation techniques to enhance the dataset before model training. Five distinct classifiers are developed using the transfer learning models, employing fine-tuning methods and hyperparameters to achieve exceptional accuracy. The top three classifiers are combined using the max voting ensemble method. The proposed ensemble model demonstrates significant potential in enhancing classification with an accuracy, precision, recall, f1-score of 97.50%, 96.33%, 98.67%, and 97.49% respectively.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Malaria red blood cell detection, Deep learning, Transfer Learning, Data Augmentation, Ensemble Learning |
| Subjects: | Q Science > Q Science (General) Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology |
| Divisions: | Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science |
| Depositing User: | AKINWUMI |
| Date Deposited: | 03 Oct 2023 11:30 |
| Last Modified: | 03 Oct 2023 11:30 |
| URI: | http://eprints.covenantuniversity.edu.ng/id/eprint/17327 |
Actions (login required)
 |
View Item |
