Reproduce “Reinforcement Learning for Controller Parameters in the Auto-Bidding System at Pinterest” 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
Yichu Zhou, Qinyi Chen, Fei Feng, Jacob Loh, Mehdi Ben Ayed, Lin Yang, Andrei Curelea, Jiajing Xu
RLBP uses reinforcement learning to tune bounded settings of an existing PID bidder daily, while the PID controller continues frequent bid adjustments. Its learned layer balances budget utilization and click cost as priorities shift with remaining lifetime. In a seven-day live A/B test, all-campaign eCPC fell 0.25% (p < 0.05), while revenue was unchanged; lifetime-ended campaigns instead saw eCPC rise 0.18%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.