Software implementation of the architecture presented in the paper Learning representations of multivariate time series with missing data by F. M. Bianchi et al.

TL;DR

The proposed approach allows for learning representations of real-valued time series with missing data by means of an autoencoder (AE) regularized by the Time series Cluster Kernel (TCK). In particular, the inner products of the vectorial representations generated by the AE are aligned with the entries of TCK.

Citation

Please, consider citing our paper if you are using it in your work.

@article{BIANCHI2019106973,
  title = {Learning representations of multivariate time series with missing data},
  journal = {Pattern Recognition},
  volume = {96},
  pages = {106973},
  year = {2019},
  issn = {0031-3203},
  author = {Filippo Maria Bianchi and Lorenzo Livi and Karl Øyvind Mikalsen and Michael Kampffmeyer and Robert Jenssen},
}