Skip to content

Conditional Wasserstein GAN for synthetic time series genetaion

Notifications You must be signed in to change notification settings

avkhimen/CWGAN-GP-time-series

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Conditional Wasserstein GAN for Synthetic Time Series Generation

Conditional Wasserstein GAN with gradient penalty for the generation of synthetic time series.

Description

Propriatary dataset was used to train the conditional WGAN with gradient penalty. Any dataset with shape (num_samples, num_features) will work.

The code was based on generating synthetic images, so whenever word 'images' appears, it should be interpreted as 'time-series' instead. I will fix this soon.

Run the cells in sequence in cwgan-gp_time_series.ipynb jupyter notebook. Final cell contains code to create synthetic image conditioned on a label.

Code in part based on:

Getting Started

Dependencies

  • tensorflow
  • numpy
  • matplotlib

See requirements.txt file.

License

Free to use for any purpose

About

Conditional Wasserstein GAN for synthetic time series genetaion

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published