Federated Machine Learning

Federated machine learning promises to be a cornerstone technology in Decentralized AI.

Federated learning (FedML) is used when multiple organizations (cross-silo) or edge devices (cross-device) want to collectively train machine learning models but datasets cannot be centralized because of privacy reasons (private, sensitive, regulated data) or for practical reasons (big and fast data at the edge). Instead, completely local model updates are computed and combined into a global aggregated model (Fig 1).

Fig 1. In federated learning, multiple clients, each operating independently, computes model updates using private local data. A coordinator, or server, is responsible for aggregating model updates. The coordinator is responsible for distributing the code to be executed, but has (and should not have) any administrative control over clients. This is a key pattern in decentralized AI. 

FedML provides strong input privacy by enabling training of machine learning models without moving and disclosing data. Much research is currently invested into security and FedML, for example to guard against attacks on the training data itself with the purpose of biasing the model, so called data-poisoning attacks, or reverse-engineering of training data using predictions from the federated ML model, so called inference attacks. These are general problems in adversarial ML and can be mitigated using general approaches also in a federated setting.

Read our paper for a deep-dive into scalable federated learning with our open source framework FEDn.

Read paper

Andreas Hellander explains the advantages of federated machine learning and what problems it solves.

YouTube Channel

Get started with Scaleout Studio

Sign up for early access!

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.