I have a project on cyberattack detection, but the dataset is very imbalanced so I want to use 3 different GAN techniques (GAN, WCGAN, DRAGAN) to generate the data of the minority classes and then build an LSTM model for classification.
- Generate the data of the minority classes using 3 different GAN architectures (Traditional GAN, WCGAN, DRAGAN)
- Build an LSTM classifier with each generated data
- Compare the accuracy and plot ROC Curve
Here is the data:
[login to view URL]
Hi! I have experience with GANs - I worked with RaGAN to generate the stock market data. I also used LSTM in my projects before. I use Pytorch for any NN implementations. It's not quite clear from the data which field Daha Fazla