From 13 - 15 December, the Third IEEE International Conference on Trust, Privacy and Security in Intelligent Systems, and Applications took place virtually.
The "IEEE TPS-ISA" is an international, multidisciplinary forum for the presentation of state-of-the-art innovations and discussion between academic, industrial researchers and practitioners on issues related to trust, privacy and security in emerging intelligent systems and applications.
Leibniz AI Lab researchers Iyiola Emmanuel Olatunji, Wolfgang Nejdl and Megha Khosla submitted a paper titled: "Membership Inference Attack on Graph Neural Networks". The paper was awarded the Best Student Paper Award.
In this paper, the researchers investigate the vulnerabilities of so-called Graph Neural Networks (GNNs) to privacy threats and show that even black-box access to GNNs can reveal training data information. They also propose some practical defences against such privacy attacks.
Congratulations on this award!
Link to the Paper: https://arxiv.org/abs/2101.06570