Skip to content
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.

Sign up for GitHub

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

Jump to bottom

solution to learned data augmentation. #40

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

solution to learned data augmentation. #40

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions _requests_for_research/learned-data-augmentation.html
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,4 @@
<p>⭐⭐⭐ <strong>Learned Data Augmentation.</strong> You could use a learned <a href="https://jaan.io/what-is-variational-autoencoder-vae-tutorial/" target="_blank" rel="noopener">VAE</a> of data, to perform “learned data augmentation”. One would first train a VAE on input data, then each training point would be transformed by encoding to a latent space, then applying a simple (e.g. Gaussian) perturbation in latent space, then decoding back to observed space. Could we use such an approach to obtain improved generalization? A potential benefit of such data augmentation is that it could include many nonlinear transformations like viewpoint changes and changes in scene lightning. Can we approximate the set of transformations to which the label is invariant? Check out the <a href="https://arxiv.org/abs/1611.01331" target="_blank" rel="noopener">existing</a> <a href="https://arxiv.org/abs/1702.05538" target="_blank" rel="noopener">work</a> <a href="https://arxiv.org/abs/1709.01643" target="_blank" rel="noopener">on</a> <a href="https://arxiv.org/abs/1711.04340" target="_blank" rel="noopener">this</a> <a href="https://arxiv.org/abs/1711.00648" target="_blank" rel="noopener">topic</a> <a href="http://cs231n.stanford.edu/reports/2017/pdfs/300.pdf" target="_blank" rel="noopener">if</a> <a href="https://arxiv.org/abs/1710.10564" target="_blank" rel="noopener">you</a> <a href="https://papers.nips.cc/paper/7278-learning-to-model-the-tail" target="_blank" rel="noopener">want</a> a place to get&nbsp;started.</p>

<h3>Solutions</h3>
<p> Preliminary paper describing learned data augmentation can be found <a href="https://github.com/chamorajg/learned-data-augmentation/blob/main/procedure/learned_data_augmentation.pdf"> here </a>, with the implementation avaiable <a href="https://github.com/chamorajg/learned-data-augmentation"> at this repo </a>. Current results improve the generalization of the VAE when tested on the CIFAR-10 dataset.</p>