The National Science Foundation awarded Marquette University a $173,184 Project Grant under the Computer and Information Science and Engineering program (CFDA 47.070) for the period of June 1, 2022 through May 31, 2024.
The grant funds research to develop Coordinate-Based Neural Networks (CBNNs) for improving the accuracy of computational imaging techniques used in applications across science and engineering. Specifically, the University will conduct work to establish a sampling theory for uniquely identifying CBNNs from finite sets of linear projection measurements, develop recovery guarantees for the associated non-convex optimization problem, and derive efficient algorithms for accelerating CBNN training to scale to practical imaging scenarios. The CBNN approach represents images as continuous domain functions mapped to intensities, allowing for more accurate implementation of common imaging forward models like continuous Radon and Fourier transforms. The theory and algorithms will be demonstrated on large-scale applications using real data from magnetic resonance imaging and computed tomography.
Generated 1/6/24, 12:42 PM