Uniform-Design Subsampling for Compute-Budgeted Double Machine Learning
计算预算约束下双重机器学习的均匀设计子抽样
Double machine learning (DML) combines orthogonal scores with flexible nuisance estimation, but repeated cross-fitting and repeated analysis can make full-data workflows expensive. When a fixed computational budget requires a working sample of size $r\ll n$, simple random subsampling may cover the covariate space poorly and produce unstable treated--control composition. We propose Uniform Design Double Machine Learning (UD-DML), which maps a common low-discrepancy skeleton…