Define imputation module based on GRU-D

Reference : Code of Re-implementing the methodology in the paper GRU-D

GRU-D : Recurrent Neural Networks for Multivariate Time Series with Missing Values

This method can be used for dealing with time series with missing values, especially for time series with non-fixed time intervals.

Environment

• Python 3.6.1
• PyTorch 0.4.1

Cite

@article{che2018recurrent,
  title={Recurrent neural networks for multivariate time series with missing values},
  author={Che, Zhengping and Purushotham, Sanjay and Cho, Kyunghyun and Sontag, David and Liu, Yan},
  journal={Scientific reports},
  volume={8},
  number={1},
  pages={6085},
  year={2018},
  publisher={Nature Publishing Group}
}