Reproduce “Stochastic Collapsed Variational Bayesian Inference for Latent Dirichlet Allocation” 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

James Foulds, Levi Boyles, Christopher Dubois, Padhraic Smyth, Max Welling

SCVB0 learns topics incrementally by updating word-topic assignments and aggregate counts, rather than storing a distribution for every token. It marginalizes topic and document-topic parameters to bring collapsed inference into stochastic learning. It improves held-out performance over SVB on two of three large-corpus tests and yields fewer human-judged incoherent words in small-corpus studies, within those tested settings.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.