Reproduce “Managing App Install Ad Campaigns in RTB: A Q-Learning Approach” 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
Anit Kumar Sahu, Shaunak Mishra, Narayan Bhamidipati
A Q-learning algorithm is presented for managing mobile app install ad campaigns in real-time bidding, effectively balancing bidder profit and advertiser campaign efficiency despite significant delays in conversion feedback. The method demonstrated a simultaneous achievement of a 12.18% increase in happy campaigns and a 272.07% increase in margin with a balanced objective setting.
Key insights
- A quantifiable trade-off exists between maximizing bidder profit and achieving high advertiser campaign efficiency, which can be strategically navigated by adjusting objective weighting.
- The proposed Q-learning framework successfully learns effective policies for joint optimization in complex RTB environments, overcoming the challenge of delayed and sparse feedback.
- Campaigns operating across a wider range of publishers exhibit greater 'controllability,' making it easier to steer them towards desired efficiency targets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.