Name of the participant: Daniel Zügner
Description of the IT-research project: Graph neural networks (GNNs) have transferred the potential of Deep Learning to the graph domain. Because graphs are central to many important applications, GNNs are considered an important class of models in machine learning (ML). Unlike traditional ML for text or images, ML for graphs does not consider individual data points in isolation; instead, a data point is considered together with its links to other data. Following the adage “like goes with like”, GNNs can even make predictions for nodes based only on its neighbours in the graph and without considering its properties.
Despite the recent breakthrough in network mining by GNNs, there are some obstacles that stand in the way of the widespread use of Deep Learning for graphs in the real world, especially for sensitive areas such as medicine or autonomous driving.
The goal of this project is to develop reliable GNNs that are more robust to anomalies in the data and also produce meaningful estimates of the uncertainty in the predictions.
Software Campus partners: TU München, Huawei
Implementation period: 01.02.2021 bis 30.04.2022