自动调整推荐系统参数，确保多目标约束达标
ACT: Automated Constraint Targeting for Multi-Objective Recommender Systems
- 基于离线数据对偶评估，自动识别最小参数调整集
- 可持续重训练以适应系统与用户行为变化
- 适合需要多目标平衡的生产级推荐系统
推荐系统常需在提升主目标的同时，确保次级目标满足最低阈值（即“护栏”）。这对维持一致的用户体验和平台生态至关重要，但面对系统与用户行为的独立变化，强制执行这些护栏极具挑战性，通常依赖手动超参数调优。本文提出自动化约束靶向（ACT）框架，能自动找到满足护栏所需的最小超参数调整集。ACT通过在无偏数据上进行离线成对评估来寻找解决方案，并持续重训练以适应系统与用户行为的变化。我们实证验证了其有效性，并描述了其在大规模生产环境中的部署情况。
原文摘要 · Abstract (English)
Recommender systems often must maximize a primary objective while ensuring secondary ones satisfy minimum thresholds, or "guardrails." This is critical for maintaining a consistent user experience and platform ecosystem, but enforcing these guardrails despite orthogonal system changes is challenging and often requires manual hyperparameter tuning. We introduce the Automated Constraint Targeting (ACT) framework, which automatically finds the minimal set of hyperparameter changes needed to satisfy these guardrails. ACT uses an offline pairwise evaluation on unbiased data to find solutions and continuously retrains to adapt to system and user behavior changes. We empirically demonstrate its efficacy and describe its deployment in a large-scale production environment.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。