Missing Pattern Recognized Diffusion Imputation Model for Missing Not At Random
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866910254174830592 |
|---|---|
| author | Sim, Gyuwon Lee, Sumin Bae, Heesun Na, Byeonghu Kwon, Doyun Hwang, Ju-Hee Lim, Jae-Young Moon, Il-Chul |
| author_facet | Sim, Gyuwon Lee, Sumin Bae, Heesun Na, Byeonghu Kwon, Doyun Hwang, Ju-Hee Lim, Jae-Young Moon, Il-Chul |
| contents | Missing data frequently arises across diverse domains, including time-series and image domains. In the real world, missing occurrences often depend on the unobservable values themselves, which are referred to as Missing Not at Random (MNAR). In this work, we introduce the Missing Pattern Recognized Diffusion Imputation Model (PRDIM), a novel framework that explicitly captures the missing pattern and precisely imputes unobserved values. PRDIM iteratively maximizes the likelihood of the joint distribution for observed values and missing mask under an Expectation-Maximization (EM) algorithm. In this sense, we first employ a pattern recognizer, which approximates the underlying missing pattern and provides guidance during every inference toward more plausible imputations with respect to the missing information. Through extensive experiments, we demonstrate that PRDIM consistently achieves strong imputation performance under MNAR settings across multiple data modalities. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_25439 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Missing Pattern Recognized Diffusion Imputation Model for Missing Not At Random Sim, Gyuwon Lee, Sumin Bae, Heesun Na, Byeonghu Kwon, Doyun Hwang, Ju-Hee Lim, Jae-Young Moon, Il-Chul Machine Learning Missing data frequently arises across diverse domains, including time-series and image domains. In the real world, missing occurrences often depend on the unobservable values themselves, which are referred to as Missing Not at Random (MNAR). In this work, we introduce the Missing Pattern Recognized Diffusion Imputation Model (PRDIM), a novel framework that explicitly captures the missing pattern and precisely imputes unobserved values. PRDIM iteratively maximizes the likelihood of the joint distribution for observed values and missing mask under an Expectation-Maximization (EM) algorithm. In this sense, we first employ a pattern recognizer, which approximates the underlying missing pattern and provides guidance during every inference toward more plausible imputations with respect to the missing information. Through extensive experiments, we demonstrate that PRDIM consistently achieves strong imputation performance under MNAR settings across multiple data modalities. |
| title | Missing Pattern Recognized Diffusion Imputation Model for Missing Not At Random |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2605.25439 |