Federated learning to reduce data transfers and ensure data economy

Context

In federated learning, neural networks are not trained at a central location, but distributed across individual systems. The trained networks are then merged. This has advantages in terms of data protection and data economy. But it also saves resources for data transmission and storage. It is necessary to investigate to what extent distributed learning systems can prove themselves in practice and what savings can be realized.

Tasks

• Exploration of federated learning approaches
• Transfer of a federated learning method into practice
• Evaluation of performance in practical use

Requirements

• You can program in Python.
• You have experience in the field of data science.
• You have a very good command of written and spoken German and English.