Select all correct statements from below.

Question $9$ options:

The gradient descent algorithm can be used to learn a Gaussian Mixture Model.

The log likelihood value of the Expectation Maximization $($EM$)$ algorithm always decreases with each iteration

One of the problems faced when finding a Mixture of Gaussians distribution is the presence of singularities$.$

The Identifiability problem is a major issue faced when estimating the distribution using Mixture of Gaussians method.