Updated 2022-05-22 0
Viewed 4 times

I try to fit MNIST data sets using Gaussian mixture and EM algorithm. However, because the dimension is too large, the calculation result of multivariate Gaussian distribution is always 0.How can I solve this problem?

    def GS(self, x, mean, cov):
        result = multivariate_normal(mean, cov).pdf(x)
        return result

I used SciPy's own function for calculation

🔴 No definitive solution yet