Research map / Reinforcement learning / Offline RL
Enhancing Generative Auto-bidding with Offline Reward Evaluation and Policy Search
Zhiyu Mou, Yiqin Lv, Miao Xu, Cheems Wang, Yixiu Mao, Jinghao Chen, Qichen Ye, Chao Li, Rongquan Bai, Chuan Yu, Jian Xu, Bo Zheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the KL constraint (ablation) | AIGB-Pearl · Real-world A/B ablation (8 days) | GMV: 30,906,963 -> 31,243,688 (+1.09%) | Improved |
| add the Lipschitz constraint (ablation) | AIGB-Pearl · Real-world A/B ablation (8 days) | GMV: 32,284,972 -> 32,869,329 (+1.81%) | Improved |
| add the Lipschitz constraint (ablation) | AIGB-Pearl · Real-world A/B ablation (8 days) | BuyCnt: 268,551 -> 281,979 (+0.50%) | Improved |
| add the KL constraint (ablation) | AIGB-Pearl · Real-world A/B ablation (8 days) | ROI: 4.25 -> 4.26 | No clear effect |
| AIGB-Pearl instead of DiffBid | auto-bidding · Simulated, budget 1.5k | GMV: 480.76 -> 502.98 (+4.62%) | Improved |
| AIGB-Pearl instead of DiffBid | auto-bidding · Simulated, budget 2.0k | GMV: 511.17 -> 521.84 (+2.09%) | Improved |
| AIGB-Pearl instead of DiffBid | auto-bidding · Simulated, budget 3.0k | GMV: 556.32 -> 574.17 (+3.21%) | Improved |
| AIGB-Pearl instead of DiffBid | auto-bidding · Real-world A/B (6k advertisers) | GMV: 76,390,174 -> 78,676,009 (+3.00%) | Improved |
| AIGB-Pearl instead of USCB | auto-bidding · Real-world A/B (6k advertisers) | ROI: 4.92 -> 5.13 (+4.24%) | Improved |
| AIGB-Pearl instead of MOPO | auto-bidding · Real-world A/B (6k advertisers) | ROI: 3.08 -> 3.23 (+4.87%) | Improved |
| AIGB-Pearl instead of DT | auto-bidding · Real-world A/B (6k advertisers) | GMV: 34,808,665 -> 35,957,933 (+3.30%) | Improved |
| AIGB-Pearl instead of DiffBid on unseen advertisers | auto-bidding · Real-world A/B, unseen advertisers | GMV: 67,092,973 -> 69,252,539 (+3.32%) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MOPO — MOPO: Model-based Offline Policy Optimization (NeurIPS 2020) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GenAuction: A Generative Auction for Online Advertising (AAAI 2025 · similarity 0.87)
- AuctionNet: A Novel Benchmark for Decision-Making in Large-Scale Games (NeurIPS 2024 · similarity 0.87)
- LLM-Auction: Generative Auction towards LLM-Native Advertising (NeurIPS 2026 · similarity 0.86)
- No-Regret Online Autobidding Algorithms in First-price Auctions (NeurIPS 2025 · similarity 0.84)
- Learning Personalized Ad Impact via Contextual Reinforcement Learning under Delayed Rewards (NeurIPS 2025 · similarity 0.84)
- Improving Reward Model Generalization from Adversarial Process Enhanced Preferences (ICML 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
