Privacy and Fairness

Description

As a result of digitalisation, we are now surrounded in our everyday lives by a large number of complex systems and services that store our data electronically and process it in some form or another. This development offers many advantages, but also poses serious risks in terms of system security and the protection of users' privacy. There have been numerous known attacks on such systems in which private user data has been stolen.

The principle of “privacy by design” means that, from the design stage of a system, special care is taken into account of data protection measures and to collect only that type of user data that is actually necessary for the correct functioning of the system. By using cryptographic methods, it is often possible to further reduce the amount of data that a system operator actually learns without affecting the system's functionality. For example, computations on sensitive user data can be performed by the user himself, so that the system operator never sees the data. With the help of so-called “zero-knowledge proofs”, the user can then convince the system operator that he has performed the computation correctly. “Multi-Party Computation” is also a popular way to protect user data. Here, several parties jointly perform a calculation on all their private data without the other parties learning information about their own data. Closely related to this is “Secure Function Evaluation”, where one party is in possession of secret data and the other party is in possession of a secret function. Now both parties can apply the function to the data together without learning the secret of the other party.

The topic of “fairness” also plays a role in the digitalised world. In the future it could become common practice to have decisions made directly by intelligent algorithms without human influence. It is important to make sure that these algorithms act “fair”, i.e. that they do not disadvantage anyone through their decisions. Furthermore, fairness is a topic in the field of “multi-party computation“. This ensures that either all parties learn the result of the calculation or nobody does.

Our research

Our teaching