FedIA

Traditional learning models require larger and more diversified data for better training. However, a particularity in the context of medical data is that data sharing between different institutions is often complicated by strict confidentiality regulations and data ownership issues, making it practically impossible to collect diversified centralized datasets on a large scale. Therefore, this project aims to develop a distributed dataset training tool without sharing data and violating privacy and ownership restrictions, using federated learning, enabling an individual medical center to benefit from the rich datasets of multiple centers without centralizing data in one place in order to perform better patient monitoring and disease diagnosis.