RSA-Regression
Regularization Self-Attention Regression
Installation
- Python 2.7
- Tensorflow-gpu 1.5.0
- Keras 2.1.3
- scikit-learn 0.19
Train the model
Run command below to train the model:
- Train RSA-Regression model based on Gold-price dataset or Palladium-price datset.
python RSA-Regression.py
You can choose different datasets. Just change the dataset path.
- Train the baseline models based on Gold-price dataset or Palladium-price datset. For example, you can choose LSTM.
python LSTM.py
Parameter study
Run command below to investigate the parameters:
- Investigate the impacts of parameters based on Gold-price dataset. For example, you can investigate the impact of number of CNN filters.
python RSA-nfilters.py
Experiment
Data are obtained from Macrotrend.
device: GTX 1050
OS: Ubuntu 16.04
dataset: Gold-price and Palladium-price
Citation
@ARTICLE{8943215,
author={J. {Zhou} and Z. {He} and Y. N. {Song} and H. {Wang} and X. {Yang} and W. {Lian} and H. {Dai}},
journal={IEEE Access},
title={Precious Metal Price Prediction Based on Deep Regularization Self-Attention Regression},
year={2020},
volume={8},
number={},
pages={2178-2187},}