Smashing the Data Bottleneck with Federated Machine Learning on the BEAM

If you use a centralized, batch-based approach to data management and analysis, the cost to derive intelligence from that data using machine learning techniques could be staggering. Massive amounts of data must be stored, protected, and analyzed, all before even machine learning techniques are applied. This process is slow and less secure. Federated learning is a type of distributed machine learning that provides a faster, less expensive and more secure solution.

In this talk, we will learn about various challenges of federated learning and walk through a code tutorial to see how Elixir, Erlang and BEAM help solve some of those challenges.

OBJECTIVES

The main objective of the talk is to demonstrate how Elixir, Erlang and BEAM help solve some of the challenges of Federated Machine Learning. We will then briefly examine a mind-map summary of a survey of federated learning system development and identify some references to guide future work on solving other open challenges.

AUDIENCE

Anyone interested in grasping the core concepts of federated machine learning and seeing how it can be implemented on the BEAM. Working knowledge of machine learning might help to get a deeper understanding.