search

WenjieDu / PyPOTS

A Python toolkit/library for reality-centric machine/deep learning and data mining on partially-observed time series, including SOTA neural network models for scientific analysis tasks of imputation, classification, clustering, forecasting, & anomaly detection on incomplete industrial (irregularly-sampled) multivariate TS with NaN missing values
https://pypots.com
BSD 3-Clause "New" or "Revised" License
818 stars 82 forks source link

Add ImputeFormer #451

Open WenjieDu opened 1 week ago

WenjieDu commented 1 week ago

1. Model description

ImputeFormer should be included in PyPOTS as an imputation model.

@article{nie2023imputeformer,
  title={ImputeFormer: Low Rankness-Induced Transformers for Generalizable Spatiotemporal Imputation},
  author={Nie, Tong and Qin, Guoyang and Ma, Wei and Mei, Yuewen and Sun, Jian},
  journal={arXiv preprint arXiv:2312.01728},
  year={2023}
}

2. Check open-source status

3. Provide useful information for the implementation

https://github.com/tongnie/ImputeFormer

WenjieDu commented 4 days ago

We are currently using SAITS loss to train ImputeFormer in PyPOTS. Tong @tongnie will make a PR to apply the original loss function of ImputeFormer soon.