Reproduce “Local Fisher Information Enables Sparse Causal Discovery” 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
Byeongguk Kang, Donghyeon Lee, Euijong Song, Gunwoong Park
FiCS orders variables from sources and tests small sets of selected variables for whether they account for each node’s causes. It uses local Fisher information for ordering and parent selection, so estimation uses small marginals. On ER3 at p = 200, n = 100, and q = 2, mean directed-edge F1 is 0.637, supporting an advantage in this tested sparse setting, not general superiority.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.