This $500,000 National Science Foundation project grant supports research and education activities at the University of Michigan from October 2022 through September 2026 under the Computer and Information Science and Engineering program.
The University of Michigan will conduct collaborative research to advance graph neural network theory, models, and applications for heterophilous data. The researchers will develop new graph neural network designs and architectures that perform well across different levels and types of heterophily, while also ensuring robustness, fairness and transparency crucial for algorithmic decision-making. In addition to advancing the theoretical understanding of heterophily in graph neural networks, the University of Michigan will explore new applications in collaboration with academic and industry partners. The award also supports training of undergraduate and graduate students at the University of Michigan, New Jersey Institute of Technology, and Michigan State University through integration into courses and research opportunities.