This is the implementation of LSTM-based Staked Autoencoder (LSTM-SAE) model
This model is mentioned in paper by the title:
Unsupervised Pre-training of a Deep LSTM-based Stacked Autoencoder for Multivariate Time Series Forecasting Problems
There are six python files and one csv file:
- pollution.csv --> Multivariate data set
- 1layer_selection.py---> this code for hyperparameter selction of the model and 1 referes to one hidden layer
- 1layer_evaluate.py ---> this code for evaluating the model and also 1 means one hidden layer
OS: Ubuntu 17.10
OS type: 64-bit
used libraries:
1- Keras (2.1.5)
2- tensorflow-gpu (1.10.0)
3- hyperopt (0.1.1)
4- pandas (0.24.2)
5- scikit-learn (0.20.3)
6- scipy (1.1.0)
7- numpy (1.15.1)
8- matplotlib (2.0.0)