Reproduce “Conditional Noise-Contrastive Estimation of Unnormalised Models” 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

Ciwan Ceylan, Michael U. Gutmann

The authors introduce conditional noise-contrastive estimation (CNCE) to fit unnormalised models without computing their normalising constant. CNCE contrasts each observation with nearby noise drawn conditionally from that point, making the unknown constant cancel. In a five-dimensional ring experiment at κ=10, CNCE has about ten times lower estimation error than NCE, but Gaussian and ICA results do not establish general superiority.

Summary by alphaXiv.

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