In addition to modeling more complicated densities, Gaussian mixture models

$($GMMs$)$ are very useful for unsupervised learning and clustering. In unsupervised learning, we seek

to find meaningful classes or clusters of the unlabeled data in the feature space.