Data Completion

Data completion aims to reconstruct missing information in datasets, a crucial preprocessing step for various applications. Current research focuses on developing sophisticated imputation models, including generative adversarial networks (GANs), diffusion models, and graph neural networks (GNNs), often combined with semi-supervised learning techniques to leverage available label information. These advanced methods are being applied across diverse fields, from engineering design and drug discovery to medical imaging and program repair, improving prediction accuracy and enabling more efficient data analysis. The ultimate goal is to enhance the reliability and utility of incomplete datasets, leading to more robust and insightful scientific discoveries and improved decision-making in practical applications.