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

The idea of SimpleShot is almost the same as that of ProtoNet #10

Open
LoveMiki opened this issue Jun 29, 2021 · 1 comment
Open

The idea of SimpleShot is almost the same as that of ProtoNet #10

LoveMiki opened this issue Jun 29, 2021 · 1 comment

Comments

@LoveMiki
Copy link

ProtoNet also uses the mean of each support class as its center and then implement Euclidean distance to train the network. The difference between SimpleShot and ProtoNet is that SimpleShot just added normalization and centering to the feature vectors before computing the Euclidean distance, is that correct?

@danoneata
Copy link

danoneata commented Nov 24, 2021

I believe the models are slightly different:

  • SimpleShot trains by using a predictive function of the form w_k f(x, θ) with w_k and θ being free parameters (see first equation in §2 from their paper);
  • ProtoNet (in the Euclidean distance scenario) trains by using a predictive function of the form 2 c_k f(x, θ) - ‖c_k‖² where only θ is a free parameter, while c_k is constrained to the centroid for class k (see equation 8 in their paper); the centroid c_k also depends on θ (see equation 1 in their paper).

I assume that the different parametrisations might yield learn different values θ for the embedding function f.

(Caveat: I'm not an expert and I've just read the SimpleShot paper.)

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