RefiDiff: Progressive Refinement Diffusion for Efficient Missing Data Imputation

TL;DR

RefiDiff introduces a progressive refinement diffusion framework that effectively imputes missing data in high-dimensional, mixed-type datasets, especially under challenging MNAR mechanisms, by combining local predictions with a novel denoising network.

Contribution

It presents RefiDiff, a novel imputation framework that integrates local predictions with a Mamba-based denoising network, enabling efficient, robust, and scalable missing data imputation without extensive tuning.

Findings

Outperforms state-of-the-art methods across various missing data scenarios.

Demonstrates strong generalization and robustness on nine real-world datasets.

Effectively handles complex MNAR missingness patterns in high-dimensional data.

Abstract

Missing values in high-dimensional, mixed-type datasets pose significant challenges for data imputation, particularly under Missing Not At Random (MNAR) mechanisms. Existing methods struggle to integrate local and global data characteristics, limiting performance in MNAR and high-dimensional settings. We propose an innovative framework, RefiDiff, combining local machine learning predictions with a novel Mamba-based denoising network efficiently capturing long-range dependencies among features and samples with low computational complexity. RefiDiff bridges the predictive and generative paradigms of imputation, leveraging pre-refinement for initial warm-up imputations and post-refinement to polish results, enhancing stability and accuracy. By encoding mixed-type data into unified tokens, RefiDiff enables robust imputation without architectural or hyperparameter tuning. RefiDiff outperforms state-of-the-art (SOTA) methods across missing-value settings, demonstrating strong performance in MNAR settings and superior out-of-sample generalization. Extensive evaluations on nine real-world datasets demonstrate its robustness, scalability, and effectiveness in handling complex missingness patterns.