Reproduce “Label Distribution Learning using the Squared Neural Family on the Probability Simplex” 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
Daokun Zhang, Russell Tsuchida, Dino Sejdinovic
SNEFY-LDL models a conditional distribution over label-share vectors, rather than returning one vector for each input. It uses a squared neural density on the probability simplex, with a closed-form normalizer and moments under specified conditions. On Natural_Scene, its feature-stratified coverage exceeded a Dirichlet baseline in one comparison, and its active-learning strategy achieved lower KL divergence; both results are benchmark-specific.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.