MIDA: Multiple Imputation Using Denoising Autoencoders

289 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Synthesizing Tabular Data using Generative Adversarial Networks (2018, methodology): “Besides statistical models generating fully synthetic data, neural models are used to impute missing values in datasets; for example, Gondara and Wang [15] uses deep de-noising autoencoders, and Yoon et al. [49] use GAN.”
Machine Learning and Integrative Analysis of Biomedical Big Data (2019, methodology): “In addition, a denoising autoencoder (DAE)-based MI (MIDA) was also proposed very recently [130].”
Deep Neural Networks and Tabular Data: A Survey (2021, background): “Three of the main purposes are data augmentation [117], data imputation (i.e., the ﬁlling of missing values) [118], [119], and rebalancing [36], [37], [120], [121].”
Flow Models for Arbitrary Conditional Likelihoods (2019, methodology): “eral real-valued data and images (for inpainting). Classic methods for imputation include k-nearest neighbors (Troyanskaya et al. 2001), random forest (Stekhoven and B¨uhlmann 2011) and auto-encoder (Gondara and Wang 2018) approaches. Often, these methods infer just one single prediction for each mi”