Enhancing Generative Auto-bidding with Offline Reward Evaluation and Policy Search
Researchers have developed AIGB-Pearl, a new method to enhance AI-Generated Bidding (AIGB) for advertising. This approach integrates generative planning with policy optimization, addressing the performance limitations of existing AIGB methods that are confined to static datasets. AIGB-Pearl utilizes a trajectory evaluator to gauge the quality of generated scores and a KL-Lipschitz-constrained score-maximization scheme for safe exploration beyond offline data. Experiments on simulated and real-world advertising systems show that AIGB-Pearl achieves state-of-the-art performance. AI
IMPACT Enhances AI-driven advertising by enabling exploration beyond static datasets, potentially improving campaign performance.