HIMCE: High-dimensional multiple imputation via covariance-mode updating for neuroimaging and spatiotemporal blocks Journal: Journal of Multivariate Analysis

Hsin-Hsiung Huang, Stef van Buuren · 2026

Read HIMCE: High-dimensional multiple imputation via covariance-mode updating for neuroimaging and spatiotemporal blocks Journal: Journal of Multivariate Analysis by Hsin-Hsiung Huang, Stef van Buuren (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-04440.txt.

Description

Hsin-Hsiung Huang and Stef van Buuren propose a faster way to fill in missing values in very high-dimensional data, such as brain imaging blocks or sensor records with dropouts. Multiple imputation is the principled approach, since it carries uncertainty about the missing values into later analysis.

Standard chained-equation methods can be slow or unstable when blocks are large and strongly correlated. An exact joint normal method becomes expensive for a different reason. It repeatedly samples and factorises a huge covariance matrix. Their hybrid keeps the same conditional imputation law but updates covariance uncertainty through its dominant mode, with stochastic draws preserving uncertainty about the mean.

Catalog record

CREATORHsin-Hsiung Huang, Stef van Buuren
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE61 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.04440

Browse the full library · About the Tower of Babel