WenjieDu/SAITS
A PyTorch-based deep learning model that uses self-attention to impute missing values in multivariate time series.
Not currently ranked — collecting fresh signals.
star history
SAITS is a PyTorch implementation of a deep learning model for time-series imputation, using self-attention mechanisms to reconstruct incomplete multivariate time series containing NaN values. The model applies dual self-attention blocks to jointly capture temporal dependencies and feature correlations for imputation tasks. The repository includes training utilities, benchmark scripts, and integrations with other time-series models via the PyPOTS framework.
Frequently asked
- What is WenjieDu/SAITS?
- A PyTorch-based deep learning model that uses self-attention to impute missing values in multivariate time series.
- Is SAITS open source?
- Yes — WenjieDu/SAITS is open source, released under the MIT license.
- What language is SAITS written in?
- WenjieDu/SAITS is primarily written in Python.
- How popular is SAITS?
- WenjieDu/SAITS has 511 stars on GitHub.
- Where can I find SAITS?
- WenjieDu/SAITS is on GitHub at https://github.com/WenjieDu/SAITS.