Showing 1–1 of 1 results for author: Roskams-Hieter, B

Leveraging variational autoencoders for multiple data imputation

Authors: Breeshey Roskams-Hieter, Jude Wells, Sara Wade

Abstract: Missing data persists as a major barrier to data analysis across numerous applications. Recently, deep generative models have been used for imputation of missing data, motivated by their ability to capture highly non-linear and complex relationships in the data. In this work, we investigate the ability of deep models, namely variational autoencoders (VAEs), to account for uncertainty in missing da… ▽ More Missing data persists as a major barrier to data analysis across numerous applications. Recently, deep generative models have been used for imputation of missing data, motivated by their ability to capture highly non-linear and complex relationships in the data. In this work, we investigate the ability of deep models, namely variational autoencoders (VAEs), to account for uncertainty in missing data through multiple imputation strategies. We find that VAEs provide poor empirical coverage of missing data, with underestimation and overconfident imputations, particularly for more extreme missing data values. To overcome this, we employ $βべーた$-VAEs, which viewed from a generalized Bayes framework, provide robustness to model misspecification. Assigning a good value of $βべーた$ is critical for uncertainty calibration and we demonstrate how this can be achieved using cross-validation. In downstream tasks, we show how multiple imputation with $βべーた$-VAEs can avoid false discoveries that arise as artefacts of imputation. △ Less

Submitted 30 September, 2022; originally announced September 2022.

Comments: 17 pages, 3 main figures, 6 supplementary figures