Privacy-friendly federated learning is primarily used in the areas of artificial intelligence, big data and smart data as well as cybercrime and cybersecurity. This is a modern method that enables artificial intelligence (AI) to learn from many different types of data without having to collect this data in a centralised location.
Instead of storing all data in a large database, as was previously the case, privacy-friendly federated learning keeps the information where it is generated - for example on users' smartphones or in individual companies. The AI learns from these different sources by exchanging secure intermediate results, known as updates. This means that sensitive data such as personal photos or health data does not end up on external servers in the first place and remains better protected.
An illustrative example: many fitness tracker users want to contribute their data to improve the training apps, but want to maintain their privacy. With privacy-friendly federated learning, the AI learns from the experience gathered on each individual tracker, but the actual data never leaves the device. In the end, everyone benefits from better suggestions and tips without revealing any personal information.
Deutsch 
English (UK)