Final master thesis supervision: Comparison of distributed machine learning techniques applied to openly available medical data

Title of the master thesis supervised: Comparison of distributed machine learning techniques applied to openly available medical data.

Author: Marco Antonio Melgarejo Aragón.

Directors: Judith Sáinz-Pardo Díaz and Álvaro López García.

Abstract: Distributed machine/deep learning refers to algorithms and systems designed to enhance performance, preserve privacy, and scale to larger training data and models. The aim of this study is to compare the performance of different distributed machine learning techniques, such as federated learning, gossip learning, or ring all-reduce architecture. To achieve this, their application is proposed using artificial neural networks on an openly available medical dataset. Various metrics will be evaluated based on the architecture configuration and the number of rounds carried out. The implementation of the three architectures using Python is proposed in a scenario where data distribution is simulated. All implemented code can be openly accessed.