• deep learning
• erlang
• federated learning
• machine learning
• middleware

Federated learning algorithms are gaining increasing interest, and their effective exploitation asks for a flexible yet efficent support to the required distributed computations. Moreover, the availability of off-the-shelf FL implementations of main Data Mining algorithms is crucial for the success of the supporting platform. The thesis work proceeded along these two development directions: improvement of the required middleware support, and coding of FL adaptations of Data Mining algorithms to be directly used on top of the middleware.