The first one is about the assignment process.
In the paper, it says that the class centers are updated in a way of the weighted moving average.
However, if we softly assign the target center e.g., 0.9 c1 + 0.1 c2, isn't the property of uniformity violated?
Or do you assign "c1" if the weight for c1 is 0.9?
How does Eq. 1 satisfy the uniformity?
If the feature dimension is 2048 and there are 1000 classes, making 1000 centers orthogonal does not seem to make them uniformly distributed.
Thanks for the interesting paper.
I have 2 questions.
However, if we softly assign the target center e.g., 0.9 c1 + 0.1 c2, isn't the property of uniformity violated? Or do you assign "c1" if the weight for c1 is 0.9?
Thank you.