Question about Equ. (2)

[yuan-zm]

Hi @theo2021 ,

For Eq. 2, I want to know why you use the global variance rather than the variance of each prototype. I.e., .sum(axis=0) in below function.

` def global_var(self):

    global_squared_mean = (
        self.squared_mean.T * self.counter / self.counter.sum()
    ).T.sum(axis=0)

    global_mean = (self.prototypes.T * self.counter / self.counter.sum()).T.sum(
        axis=0
    )

    return torch.sqrt(global_squared_mean - global_mean**2)

`