Enhancing Generative Auto-bidding with Offline Reward Evaluation and Policy Search

Researchers developed AIGB-Pearl, a new AI-driven auto-bidding system that combines generative planning with policy optimization to improve advertising performance. The system addresses limitations of existing offline reinforcement learning methods by incorporating a trajectory evaluator and safe exploration mechanisms beyond static datasets.

• →AIGB-Pearl integrates generative planning with policy optimization to enhance AI-generated bidding systems.
• →The method uses a trajectory evaluator to assess generated bid scores and ensure safe exploration beyond offline data.
• →A KL-Lipschitz-constrained score-maximization scheme enables efficient exploration while maintaining safety.
• →Testing on both simulated and real-world advertising systems shows state-of-the-art performance.
• →The approach addresses key limitations of existing offline reinforcement learning auto-bidding methods.