Hands-On Machine Learning with TensorFlow.js

Relationship with K-means

As we can infer, the estimation phase corresponds to cluster assignment in K-means, while the maximization phase corresponds to the phase where we calculate the next centroid. K-means is a hard clustering algorithm where each data point can belong to only one cluster. K-means is a special case of the EM algorithm that we described previously.

First, let's assume that each Gaussian distribution shares the same covariance matrix, expressed as . , as an identity matrix:

By using this, the responsibility of the k^th element ...