Type of machine learning algorithm used to cluster or classify data based on Gaussian (bell curve) probability distributions. GMMs are fairly quick and easy to train and perform clustering or basic classification tasks. GMMs allow “mixed membership”, so something can belong partially to one cluster or classification and another. Consider clusters as Gaussian distributions centered on the cluster centroids. K-means is actually a special case of GMM in which each cluster’s covariance along all dimensions approaches 0. This implies that a point will get assigned only to the closest cluster, while with GMM a point could be assigned to a farther cluster.