A simple and general debiased machine learning theorem with finite-sample guarantees
A simple and general debiased machine learning theorem with finite-sample guarantees
复制标题
具有有限样本保证的简单且通用的去偏机器学习定理
DOI：
10.1093/biomet/asac033
复制
发表时间：
2022
期刊：
影响因子：
2.7
通讯作者：
Singh, R
中科院分区：
文献类型：
--
作者：
Chernozhukov, V;Newey, W K;Singh, R
去偏机器学习是一种基于偏差校正和样本分割的元算法，用于计算泛函的置信区间，即，机器学习算法的标量摘要。例如，分析师可以寻求用神经网络估计的治疗效果的置信区间。我们提出了一个非渐近去偏机器学习定理，它包含了任何机器学习算法的任何全局或局部泛函，这些泛函满足一些简单的、可解释的条件。形式上，我们证明了一致性，高斯逼近和半参数有效的有限样本参数。对于全局泛函，收敛速度是，而对于局部泛函，收敛速度缓慢下降。我们的研究结果最终在一组简单的条件下，分析师可以用来将现代学习理论率转化为传统的统计推断。这些条件揭示了不适定反问题的一般双鲁棒性。
Debiased machine learning is a meta-algorithm based on bias correction and sample splitting to calculate confidence intervals for functionals, i.e., scalar summaries, of machine learning algorithms. For example, an analyst may seek the confidence interval for a treatment effect estimated with a neural network. We present a non-asymptotic debiased machine learning theorem that encompasses any global or local functional of any machine learning algorithm that satisfies a few simple, interpretable conditions. Formally, we prove consistency, Gaussian approximation and semiparametric efficiency by finite-sample arguments. The rate of convergence isfor global functionals, and it degrades gracefully for local functionals. Our results culminate in a simple set of conditions that an analyst can use to translate modern learning theory rates into traditional statistical inference. The conditions reveal a general double robustness property for ill-posed inverse problems.