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

A proposed Variational Autoencoder Hack: Focal Loss with greedy pretraining #61

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

A proposed Variational Autoencoder Hack: Focal Loss with greedy pretraining #61

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d769699
Update README.md
Jan 18, 2020
98035a0
Merge pull request #1 from PsycheShaman/PsycheShaman-patch-1
Jan 18, 2020
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
5 changes: 5 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,6 +130,11 @@ while lossG > B:
- Apply on several layers of our generator at both training and test time
- https://arxiv.org/pdf/1611.07004v1.pdf

## 18: For VAEs: Use Focal Loss as Reconstruction Error
- Greedy pretraining: use MSE loss at high learning rate (~0.01) for few epochs until output appears to be in the right "ballpark"; Tanh activation output layer; images scaled \[-1,1\]
- Subsequently, scale images between \[0,1\], loading the saved weights from greedy pretraining phase, use Sigmoid activation function in output layer and Focal loss as reconstruction error at a very low learning rate (1e-6 or less)
- See custom Tensorflow loss functions [in Python with link to paper on arXiv](https://github.com/umbertogriffo/focal-loss-keras) and [in R](https://gist.github.com/PsycheShaman/ea39081d9f549ac410a3a8ea942a072b)


## Authors
- Soumith Chintala
Expand Down