Dictionary Learning from joint time serieses using Online Matrix Factorization (flexible nonnegativity constraints on dictioanry and code matrices) Learns dictionary atoms for short time-evolution patterns of multiple entities and uses them to reconstruct the time-series data
These codes are based on the following papers
-
Hanbaek Lyu, Christopher Strohmeier, Deanna Needell, and Georg Menz, “COVID-19 Time Series Prediction by Joint Dictionary Learning and Online NMF” https://arxiv.org/abs/2004.09112
-
Hanbaek Lyu, Palina Salanevich, Jacob Li, Charlotte Huang, and Deanna Needell "Temporal Dictionary Learning for EEG and Constructing Correlation Tensor" In preperation.
- utils/TDL.py : Main file implementing temporal dictionary learning
- utils/TDL_plotting.py : Helper functions for plotting
- utils/onmf.py : Online Nonnegative Matrix Factorization algorithm (generalization of onmf to the tensor setting by folding/unfolding operation)
- covid_dataprocess.py : Preprocessing functions (modify this for your own data type)
- TDL-COVID-Test.ipynb : Jupyter notebook example of temporal dictionary learning
- Hanbaek Lyu - Initial work - Website
This project is licensed under the MIT License - see the LICENSE.md file for details