Timeseries, in particular signal timeseries, are distinct in many respects - so GANs working on images may not work for timeseries. Since other questions asking on data augmentation, GANs have progressed, for example:
All above have a theme in common: images. -- This said: can GANs be used for timeseries data augmentation in 2019? If so, any examples of implementations that would work well?
I'm training a CNN-RNN EEG seizure (binary) classifier. Data specs:
- 10 minute segments sampled at 400-Hz --> 240000 timesteps per sample
- 4520 samples: 660 positive (seizure), 3860 negative (non-seizure)
- 16 channels in each sample; samples stem from 3 patients
- ~1.74E10 datapoints in total