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

Inquire about some questions about the paper "Unsupervised Video Domain Adaptation for Action Recognition: A Disentanglement Perspective" #4

Open
ztt159 opened this issue Jan 18, 2024 · 1 comment

Comments

@ztt159
Copy link

ztt159 commented Jan 18, 2024

Dear Sir, greetings. Firstly, congratulations on the successful publication of your paper in NeurIPS. However, I have some minor questions regarding its content that I would like to consult with you. Concerning the Domain Specificity & Static Consistency module, in the subsequent operations, it is mentioned that "The static latent factors disentangled from the original sequence and the shuffled sequence should be ideally equal or be very close at least. This motivates us to minimize the distance between these two static factors. Meanwhile, to further enhance the domain specificity, we enforce the dynamic latent factors from different domains to have a large distance." This is the part I find perplexing. Shouldn't we aim to maximize static latent factors? This is in consideration of your earlier statement, "the dynamic latent factors are enforced to be domain-invariant."

@wpf89928
Copy link
Collaborator

Dear Sir, greetings. Firstly, congratulations on the successful publication of your paper in NeurIPS. However, I have some minor questions regarding its content that I would like to consult with you. Concerning the Domain Specificity & Static Consistency module, in the subsequent operations, it is mentioned that "The static latent factors disentangled from the original sequence and the shuffled sequence should be ideally equal or be very close at least. This motivates us to minimize the distance between these two static factors. Meanwhile, to further enhance the domain specificity, we enforce the dynamic latent factors from different domains to have a large distance." This is the part I find perplexing. Shouldn't we aim to maximize static latent factors? This is in consideration of your earlier statement, "the dynamic latent factors are enforced to be domain-invariant."

Hi, thanks for your interest in our work. This is a typo. It should be "we enforce the static latent factors from different domains to have a large distance".

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

2 participants