Distributed Machine Learning: Improved Data Protection for AI Applications?

AI systems are based on the training of large amounts of – sometimes sensitive – data. The development of AI applications with personal data presents companies with legal uncertainties; the hurdles to compliance with data protection and the right to informational self-determination are high. A technical solution to create privacy-preserving AI applications promises the method of distributed machine learning. The data used in the training of AI algorithms is not bundled centrally, but remains on the end devices – and thus with the users. In particular, AI-based health solutions that use personalised patient data, for example to detect cases of disease such as Covid-19 or leukaemia, can benefit from the distributed machine learning method. However, new points of attack for cybercriminals are conceivable and need to be discussed.

An overview of the method and application examples of distributed machine learning is given in the first issue of AI AT A GLANCE. The new publication series of the Plattform Lernende Systeme provides concise and well-founded expertise on current developments in the field of artificial intelligence and highlights potentials, risks as well as open questions.