What is it about?
Missing data is an important problem in scientific computing. We introduce the first Transformer-based architecture that integrates the low-rank factorization into the model structure. This integration enables our model to balance between noise and signals, making it efficient for data imputation.
Featured Image
Photo by Conny Schneider on Unsplash
Read the Original
This page is a summary of: ImputeFormer: Low Rankness-Induced Transformers for Generalizable Spatiotemporal Imputation, August 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3637528.3671751.
You can read the full text:
Resources
Contributors
The following have contributed to this page
