Reproduce “Faithful Federated Tabular Data Synthesis” 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
Zilong Cao, Hai Zhang, Xuan Bi
FITS generates mixed-type synthetic tabular records from federated data without pooling raw rows. It estimates conditional distributions locally, combines client summaries in two communication rounds, and applies the probability chain rule; discrete categories are spread over continuous intervals and mapped back. In one simulation, FITS-G’s joint-distribution energy distance was 0.0429, lower than two baselines, only for the tested setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.