Project Grant 2210019

This $400,000 National Science Foundation Project Grant, awarded under the Mathematical and Physical Sciences program (CFDA 47.049), will support the development of statistical methods and machine learning techniques for analyzing complex structured and count data. Over a three-year period ending in August 2025, the University of Washington will advance the state of knowledge in big structured and count data analysis through two tracks of research. The first track will focus on revising and generalizing random graph-based statistical inference methods like nearest neighbor matching and graph-based correlation coefficients to boost efficiency while maintaining robustness and computational speed. The second track will bridge heterogeneous count-valued mixtures to nonparametric models under the umbrella of heterogeneous mixture model-based inference. In addition to furthering theory, methods, computation and applications across both tracks, preliminary results from the first track have already stimulated new work in causal inference and results from the second track are expected to help with early diagnosis of autism.