0

I'm working on a signal denoising problem. Because of not having enough data for training I'm considering using one of the generative models like VAE or GAN to generate data similar to real data for training. I didn't find any relevant precedents with regard to my use case for generative models. Is it because generative models don't work for my purpose or am I missing something?

Xiaohong Deng
  • 1
  • 1
  • I think you need to look closer to the GAN idea. This is mainly their idea with noisy image. Maybe i don't get what you want to do i suppose :( – PauZen Sep 13 '19 at 09:44
  • @PauZen Say I want to train an autonomous driving system to drive in an area where it doesn't snow much. To make the system robust, I want it to be able to drive in such weather but I don't have data in snowy days. Can I use GAN to generate fake snowy data based on real non-snowy data to train my model to drive in snowy days? My problem is similar except that the suitable generative model might be one other than GAN. I need time series data. – Xiaohong Deng Sep 14 '19 at 03:52
  • Something like this https://arxiv.org/pdf/1806.08666.pdf ? I mean using recurrent neural network (to deal with time series data). – PauZen Sep 14 '19 at 05:14

0 Answers0