Reproduce “Beyond Best-of-$k$: Prior-Consistent DAPS for Ambiguous Nonlinear 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
Merve Kocabas, Benjamin A. Burns, Andreas Geiger, Kent Gauen, Sara Fridovich-Keil
PC-DAPS addresses Fourier phase-retrieval ambiguity by keeping multiple reconstructions under one measurement and using a learned image prior to prune candidates that appear prior-inconsistent. On FFHQ, faithful outputs among four samples per image rise from 79.5% to 97.5%, while upside-down or color-inverted outputs fall from 14.5% to 1.5%. Gains are smaller on ImageNet, where the prior gives weaker disambiguating signals.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.