{"id":"arxiv/2610.23315","url":"https://arxiv.org/abs/2610.23315","tags":["cs.LG","math.OC"],"score":0,"title":"Untrained CNN and coefficient-space regularization for joint reconstruction in photoacoustic tomography","domain":"cs.LG","source":"arxiv","authors":["Alfio Bozi","Anwesa Dey","Souvik Roy"],"reviews":0,"subject":"cs.LG","summary":"The joint reconstruction of the initial pressure and spatially varying sound speed in photoacoustic tomography from boundary measurements is considered. A regularized data-driven framework is proposed in which the unknown fields are represented in finite-dimensional wavelet spaces and fixed nonlinear features generated by an untrained convolutional neural network (CNN) are used to reconstruct the initial-pressure coefficients through Tikhonov-regularized least squares. A learned coefficient-space prior is subsequently used to infer the squared sound-speed coefficients. Numerical experiments…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-10-16","reactions":{}}