Centroid-Referenced Mahalanobis Matching (CRM): A Scalable, Representation-Based Framework for Causal Inference in Large Observational Studies
质心参考马氏匹配(CRM):面向大规模观察性研究因果推断的可扩展、基于表示的框架
AI总结 针对大规模观察性研究因果推断的匹配方法存在成本高、重叠有限时改变目标总体的问题,本文提出CRM框架,其计算成本低、保留处理单元比例高、平衡表现优且速度快,贡献在于可扩展性与显式支撑诊断。
Comments Accepted by Transactions on Machine Learning Research (TMLR), 2026. Code: this https URL (https://github.com/KemingHu-D/crm-matching)