Reproduce “Generalizing Bayesian Optimization with Decision-theoretic Entropies” 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

Willie Neiswanger, Lantao Yu, Shengjia Zhao, Chenlin Meng, Stefano Ermon

The authors introduce expected H-information gain (EHIG), a Bayesian-optimization framework that selects queries by expected reduction in the best achievable posterior expected loss for a specified final decision. It recovers established acquisitions under particular actions and losses. In tested examples, HES spreads queries across five Alpine optima and leads several task-specific comparisons; these 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.