The model is a Conditional Generative Adversarial Network for time series with not regular time intervals.
The model is created to generate new time series given a training set of them.
The main idea is to use this model to augment unbalanced dataset of time series, in order to increase the precision of a classifier.
-
Requirements:
- python 3
- tensorflow, numpy
-
Download the repository
-
python3 main.py N M file_in file_times file_out
- N = training set size
- M = time series length
- file_in = file input path
- file_times = file with time stamps of the new time series
- file_out = file output path