From the course: Cloud-Based AI Solution Design Patterns

Unlock this course with a free trial

Join today to access over 24,800 courses taught by industry experts.

Federated AI learning

Federated AI learning

- What if your organization wanted to collaborate with others to develop an AI model, perhaps to be used by a community of organizations? Each organization will want to contribute their internal data to the training process, but won't want to share that data with others. The federated AI learning pattern addresses this scenario by providing a means by which each participating organization can train a version of the model using their local training data, and then provide their updates to a global model. So a typical arrangement is for there to be a global model deployed on a central server, located on a third party cloud. The initial global model is shared with all participating organizations as a starting point. One organization, let's call them organization A, uses their own training data to train a version of the model. They then share how the initial global model they received has changed by providing a set of instructions on how the global model on the central server should be…

Contents