New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Examples on Multivariate Time Series #90
Comments
This seems like a really interesting area. We have been planning to diversify this framework and this looks like an interesting direction to proceed. I could find the following papers on this topic. (Feel free to extend the list): |
For the weekend imputation problem of the dataset I have shared, one can create a training set that consists of data from Monday to Friday at each week. Then, impute the weekend before or after using the Generator. Lastly, the imputed seven-days signal can be fed to Discriminator, so that predicts whether the first-two days or last two-days of that signal was imputed. Note: One can probably impute both prior and after weekend at the same time and take the average of the losses for these predictions for both cases on each sample week. |
Dear @avik-pal & @Aniket1998, I have attempted making an implementation of what I have described as a potential solution to above mentioned weekend imputation problem. I would be more than glad if you can check #99; as I am quite unfamiliar with GAN and this was my initial attempt to employ it. |
@kayuksel Thanks for the attempt. However, we would really appreciate if you could put the code in some git repository or maybe a gist. It makes reviewing it much simpler. |
BTW I did go through your code. It seems fine. The only issue is that it is not exactly the style we follow in torchgan, you could actually use the convenience functions that we provide instead of writing everything explicitly. That being said it is not very difficult to modify it. I would recommend that you open a PR with this in the example directory as a script for now. I can then review it and suggest the changes that you can make. Overall it seems like a good demonstration for a non-CV Problems. @Aniket1998 and I have had discussions regarding expanding the base of this framework to support a wide variety of GANs and this seems to be a good start. |
The current framework is too much oriented towards Computer Vision problems where there is already lots of available resources. It would be great if you can also focus on multivariate time series where Generative Adversarial Networks can have several very important uses in data generation, augmentation, imputation, denoising and detecting anomalies for several applications such as medical, financial, etc.. It would be great if you can add examples for few of these use-cases. Enclosed is such multi-variate time series dataset consisting of log-returns where cubic interpolation is employed for imputation and market with a Boolean, I am sending it in case you would like to have an example for such a common use-case in financial world.
Dataset.zip
The text was updated successfully, but these errors were encountered: