The National Science Foundation (NSF) awarded a $400,000 Project Grant under the Computer and Information Science and Engineering (CFDA #47.070) program to the University of Illinois for a 4-year collaborative research project on privacy-preserving machine learning on graph-structured data. The project aims to develop innovative, efficient algorithms for training and updating large-scale graph neural network models while preserving the privacy of sensitive graph data across applications in areas like communication theory, computational biology, and social sciences. Key research activities include devising novel privatization protocols, implementing methods for removing graph information without retraining, and developing more reliable membership inference approaches to measure model information leakage. The project will also provide cross-disciplinary student training opportunities and increase participation of underrepresented groups in STEM research through targeted recruiting and student exchange programs.
Generated 12/31/24, 12:43 PM