MISNN: 半参数神经网络多元插补（英文版）

英文标题：MISNN: Multiple Imputation via Semi-parametric Neural Networks中文摘要：本文提出了一个新的算法 MISNN，用于处理高维缺失数据的多重插补问题，该算法借助于神经网络的逼近能力，将特征选择嵌入 MI 模型中，经过实验证明在插补准确度、统计一致性和计算速度等方面优于现有的 Bayesian Lasso 和矩阵补全等最新算法。英文摘要：Multiple imputation (MI) has been widely applied to missing value problems inbiomedical, social and econometric research, in order to avoid improperinference in the downstream data analysis. In the presence of high-dimensionaldata, imputation models that include feature selection, especially $\ell