Reproduce “Exploratory creative pre-ranking: Hierarchical empirical-Bayes Thompson sampling for real-time creative optimization” 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
Jiale Han, Ajith Moparthi, Dan Xie, Xiaowu Dai
TS-RR selects ad creatives for a personalized ranker using audience-segment estimates and Thompson sampling. It borrows evidence across a hierarchy, retains uncertainty for under-sampled creatives, and forwards a quota-limited shortlist. In a frozen replay at a global cap of 600, it improved predicted play-through rate by 8.14%, versus 2.80% for an equal-size random shortlist; this is not a measured live effect.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.