Date of Graduation
5-2023
Document Type
Thesis
Degree Name
Bachelor of Science in Computer Science
Degree Level
Undergraduate
Department
Computer Science and Computer Engineering
Advisor/Mentor
Nakarmi, Ukash
Committee Member/Reader
Gauch, Susan
Committee Member/Second Reader
Le, Thi Hoang Ngan
Abstract
The massive amount of data available in our modern world and the increase of computational efficiency and power have allowed for great advancements in several fields such as computer vision, image processing, and natural languages. At the center of these advancements lies a data-centric learning approach termed deep learning. However, in the medical field, the application of deep learning comes with many challenges. Some of the fundamental challenges are the lack of massive training datasets, unbalanced and heterogenous data between health applications and health centers, security and privacy concerns, and the high cost of wrong inference and prediction. One of the interesting questions of data-centric learning in the medical field is whether we can leverage the heterogenous data available in several medical facilities in a combined way without actually sharing the data between the institutes and preserving the security and privacy of patients. One way to address this question is through the use of the federated deep learning technique. In federated deep learning, the “learning” from each local deep learning model trained on a small, distinct dataset is shared with a global model instead of sharing the actual data and hence does not violate any security and privacy concerns.
In this study, we aim to evaluate the efficiency of the federated learning approach on classification tasks in the medical image domain. Learning in the medical image domain is often more challenging and distinct than that of natural images because of heterogeneity in the data and the unavailability of clear, discernable discriminant features between images of different classes. To this end, we investigate federated learning in medical images in terms of model architecture and data complexity. Through our experiments, we will also investigate the effect that federated learning will have on each local model’s performance, and how it affects model generality to external datasets.
Keywords
Deep learning; Federated learning; Artificial Intelligence; Privacy
Citation
Brixey, J. (2023). Analysis of a Federated Learning Framework for Heterogeneous Medical Image Data: Privacy and Performance Perspective. Computer Science and Computer Engineering Undergraduate Honors Theses Retrieved from https://scholarworks.uark.edu/csceuht/115
