提出首个针对多表数据删除的可学习基数估计框架，解决过估难题。
Forgetting by Pruning: Data Deletion in Join Cardinality Estimation
- 通过敏感度剪枝与域剪枝，精准定位并移除被删数据影响的参数
- 在高删除率下仍保持最低查询误差，优于全重训练
- 计算开销极小，仅占微调时间的0.3%-2.5%，适合生产部署
在多表关系数据中，学习型基数估计（CE）系统面临独特的机器遗忘挑战。数据删除作为核心环节，需应对属性级敏感性、表间传播及域消失导致的严重过估计问题。本文提出首个专为多表学习型CE设计的遗忘框架——基数估计剪枝（CEP）。CEP引入分布敏感剪枝，构建半连接删除结果并计算敏感度评分以指导参数剪枝；同时采用域剪枝，彻底移除被删除数据所依赖的值域支持。我们在NeuroCard和FACE两个先进架构上，基于IMDB和TPC-H数据集进行评估。结果表明，CEP在多表场景下持续取得最低的查询误差（Q-error），尤其在高删除比例下表现突出，常优于完整重训练。此外，CEP显著减少收敛迭代次数，计算开销仅为微调时间的0.3%-2.5%。
原文摘要 · Abstract (English)
Machine unlearning in learned cardinality estimation (CE) systems presents unique challenges due to the complex distributional dependencies in multi-table relational data. Specifically, data deletion, a core component of machine unlearning, faces three critical challenges in learned CE models: attribute-level sensitivity, inter-table propagation and domain disappearance leading to severe overestimation in multi-way joins. We propose Cardinality Estimation Pruning (CEP), the first unlearning framework specifically designed for multi-table learned CE systems. CEP introduces Distribution Sensitivity Pruning, which constructs semi-join deletion results and computes sensitivity scores to guide parameter pruning, and Domain Pruning, which removes support for value domains entirely eliminated by deletion. We evaluate CEP on state-of-the-art architectures NeuroCard and FACE across IMDB and TPC-H datasets. Results demonstrate CEP consistently achieves the lowest Q-error in multi-table scenarios, particularly under high deletion ratios, often outperforming full retraining. Furthermore, CEP significantly reduces convergence iterations, incurring negligible computational overhead of 0.3%-2.5% of fine-tuning time.
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。