The code for paper "End-to-End Incomplete Time Series Modeling from Linear Memory of Latent Variables" accepted by IEEE Transactions on Cybernetics.
The demo for dataset 'SCITOS G5' with 50% missing ratio
The published imputation result: 0.170
data/raw.txt: raw_data for evaluation
data/miss_data.txt: missing_data
tf_version: 1.11.0
python_version: 2.7.17
Usage: python LIMELSTM.py
@ARTICLE{Ma2019end,
author={Q. {Ma} and S. {Li} and L. {Shen} and J. {Wang} and J. {Wei} and Z. {Yu} and G. W. {Cottrell}},
journal={IEEE Transactions on Cybernetics},
title={End-to-End Incomplete Time-Series Modeling From Linear Memory of Latent Variables},
year={2019},
pages={1-13},
ISSN={2168-2267}
}