Federated learning is a term used in the fields of artificial intelligence, big data, smart data and cybersecurity. It is a special method by which many computers work together to train an artificial intelligence - without having to send their own data to a central location.
Imagine that many hospitals want to jointly develop an AI that recognises diseases on X-ray images. Instead of passing on all sensitive patient data to a centralised system, each data set remains in the respective hospital. The AI is trained on site and only what it has learnt - i.e. the improved calculation rules - is anonymised and merged into a joint model. In this way, everyone benefits from each other without anyone giving up their sensitive data.
Federated learning therefore protects private data and fulfils important data protection requirements, for example in accordance with the GDPR. This method is increasingly being used in areas such as healthcare, smartphones and connected cars. Companies can use it to work together on innovations - without disclosing their data.