University researchers rely on artificial intelligence (AI) in app development
Lübeck. Originally, the federal government wanted to launch a contact tracing app as early as mid-April in order to be able to quickly track the contacts of infected people. But a dispute over the way in which data is stored and ambiguities about competences are causing delays. Researchers from Lübeck University have now spoken out on this subject. They stress that even with a decentralized solution, such a “tracking app” can implement many possibilities for extensively evaluating the infection dynamics.
As an example of the advantages of a decentralized solution, they cite a timely analysis of the risk of infection due to a lack of distance after loosening measures. This is also possible without central algorithms for the evaluation of the data by methods of artificial intelligence (AI). In doing so, the scientists are relying on an extension of so-called Federated Learning, which already enables decentralized learning on distributed data, but largely without the necessary data protection.
Methods of privacy protection should therefore be included. In addition, the computer scientists propose to offer further functions that provide further data of the movement profile on a voluntary basis. “As the central location of the AI flagship project AI-SIGS, we are predestined to use AI in the fight against Covid-19,” says Prof. Thomas Martinetz, co-speaker of the University’s Center for Artificial Intelligence. And university vice-president Prof. Stefan Fischer emphasizes: “We consider the path taken by the Federal Government of a decentralized Corona app to be the right one, but we strongly advocate using the app even more extensively than previously planned.” mho