Reproduce “Model-Aware Regularization For Learning Approaches To Inverse Problems” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jaweria Amjad, Zhaoyan Lyu, Miguel R. D. Rodrigues
The authors introduce an inverse-problem training method that uses a known measurement operator to regularize neural-network reconstructions. It penalizes the network’s input-output Jacobian and its composition with the operator. On fastMRI knee data with a 1D mask retaining 25% of Fourier samples, it reports 29.30 dB PSNR and 0.78 SSIM, versus 28.81 dB and 0.74 for plain U-Net.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.