Generative Adversarial Imputation

Generative adversarial imputation (GAI) uses generative adversarial networks (GANs) to estimate missing values in datasets, aiming to improve data quality and enable more robust analyses. Current research focuses on enhancing GAI's performance through architectural innovations, such as incorporating graph neural networks to capture data dependencies or convolutional layers for spatio-temporal data, and by addressing limitations like high missing data rates via transfer learning. These advancements are improving imputation accuracy and efficiency across diverse applications, including engineering design, autonomous driving (e.g., pedestrian pose estimation), and scientific simulations, ultimately leading to more reliable and insightful results from incomplete data.

Papers

June 17, 2024

Bridging Design Gaps: A Parametric Data Completion Approach With Graph Guided Diffusion Models
Rui Zhou, Chenyang Yuan, Frank Permenter, Yanxia Zhang, Nikos Arechiga, Matt Klenk, Faez Ahmed
Generative Modeling Bridging Gap Data Completion Generative Adversarial Imputation Bicycle Design

June 6, 2023

SDR-GAIN: A High Real-Time Occluded Pedestrian Pose Completion Method for Autonomous Driving
Honghao Fu, Libo Sun, Yilang Shen, Yiwen Wu
Autonomous Driving 3D Human Pose Pedestrian Detection Generative Adversarial Imputation

February 6, 2023

ClueGAIN: Application of Transfer Learning On Generative Adversarial Imputation Nets (GAIN)
Simiao Zhao
Transfer Learning Application Proficiency Superficial Clue Generative Adversarial Imputation

October 31, 2022

Diffusion models for missing value imputation in tabular data
Shuhan Zheng, Nontawat Charoenphakdee
Diffusion Model Tabular Data Time Series Imputation Generative Adversarial Imputation

January 10, 2022

Differentiable and Scalable Generative Adversarial Models for Data Imputation
Yangyang Wu, Jun Wang, Xiaoye Miao, Wenjia Wang, Jianwei Yin
Data Imputation Deep Imputation Generative Adversarial Imputation

December 21, 2021

Multiple Imputation via Generative Adversarial Network for High-dimensional Blockwise Missing Value Problems
Zongyu Dai, Zhiqi Bu, Qi Long
Generative Adversarial Network Missing Value Generative Adversarial Imputation

November 3, 2021

Convolutional generative adversarial imputation networks for spatio-temporal missing data in storm surge simulations
Ehsan Adeli, Jize Zhang, Alexandros A. Taflanidis
Raw Data Deep Convolutional Neural Network Knowledge Enhanced Conditional Imputation Spatiotemporal Imputation Generative Adversarial Imputation Categorical Imputation

Generative Adversarial Imputation

Papers

Bridging Design Gaps: A Parametric Data Completion Approach With Graph Guided Diffusion Models

SDR-GAIN: A High Real-Time Occluded Pedestrian Pose Completion Method for Autonomous Driving

ClueGAIN: Application of Transfer Learning On Generative Adversarial Imputation Nets (GAIN)

Diffusion models for missing value imputation in tabular data

Differentiable and Scalable Generative Adversarial Models for Data Imputation

Multiple Imputation via Generative Adversarial Network for High-dimensional Blockwise Missing Value Problems

Convolutional generative adversarial imputation networks for spatio-temporal missing data in storm surge simulations