Reproduce “Gated Domain Units for Multi-source Domain Generalization” 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
Simon Föll, Alina Dubatovka, Eugen Ernst, Siu Lun Chau, Martin Maritsch, Patrik Okanovic, Gudrun Thäter, Joachim M. Buhmann and 2 more
The generalized domain unit layer models source and unseen domains as mixtures of shared latent distributions, then weights specialist predictions by how closely an observation matches learned units. The authors evaluate it on eight WILDS datasets. On PovertyMap, its fine-tuned MMD version reaches a worst urban/rural Pearson correlation of 0.62 (SD 0.13), above ERM and its ensemble, but results are mixed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.