federated learning

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

In federated learning many different computers work on portions of training data during machine learning and then pool parts or updates to a single resulting model. This has been applied particularly to various forms of neural network, to enhance privacy where personal data is kept in indivdiual's own devices and only weight uodates geberated by baclkprop shared back to the central model. However, it has been shown that this can be susceptable to {[adversarial attacks}}.

Defined on page 504

Used on Chap. 20: page 504