You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for the great idea of using classification layer for cosine distance computation!
Could you please elaborate on how you train that layer? As I understand you initialize weights with average feature vectors for each class and what do you do with biases? Do you use them for training?
The text was updated successfully, but these errors were encountered:
Thanks for the great idea of using classification layer for cosine distance computation!
Could you please elaborate on how you train that layer? As I understand you initialize weights with average feature vectors for each class and what do you do with biases? Do you use them for training?
The text was updated successfully, but these errors were encountered: