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.

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

Add possibility to apply transforms also to submatrices #40

Closed
einbandi opened this issue Aug 8, 2022 · 1 comment
Closed

Add possibility to apply transforms also to submatrices #40

einbandi opened this issue Aug 8, 2022 · 1 comment

Comments

@einbandi
Copy link
Collaborator

einbandi commented Aug 8, 2022

In t-SNE, for example, the probability submatrix (p_ij) needs to be normalized to make it comparable to the batch-wise probability matrix (q_ij).

I can think of two ways to add this functionality:

  • Add a list of sub_transforms to the relations class, which is only applied to submatrices.
  • Add an apply_to_sub parameter to the transforms.

I also need to check how this affects the definition of the sub method, especially if it needs to be changed to construct RelationData objects of the subsampled data first, so that the tranforms can safely be applied.

@einbandi
Copy link
Collaborator Author

einbandi commented Aug 9, 2022

Maybe the way to go is to have a normalize_rels parameter in the RelationLoss, because that's where the sub method is used.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant