Reproduce “Improving Ads Ranking Models with Auxiliary Supervision: Techniques and Insights” 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
Nikhil Bhale, Apurva Pathak, SaiSujithKumarReddy MarapaReddy, Xiao Yin, Prateek Bhansali
The authors improve post-click conversion prediction in ad ranking by training with auxiliary user-intent signals when conversion labels are missing. They describe clues and use Label Combination (LaCo) to construct auxiliary labels from signal combinations, while keeping auxiliary tasks out of inference. Offline, LaCo lowers Normalized Entropy from 0.5400 to 0.5387, a 0.25% relative improvement; online effects remain unknown.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.