Reproduce “Image segmentation with traveling waves in an exactly solvable recurrent neural network” 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
Luisa H. B. Liboni, Roberto C. Budzinski, Alexandra N. Busch, Sindy Löwe, Thomas A. Keller, Max Welling, Lyle E. Muller
The authors present a complex-valued recurrent network that segments images by encoding pixel intensity as oscillation frequency and grouping pixels with similar phase histories. Distance-dependent connections generate patterns that are clustered into object labels. On non-overlapping geometric shapes, it correctly clusters 93% of pixels across 1,000 two-shape images and 86% across 1,000 three-shape images. Natural-image evidence is limited to five examples.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.