In statistics, EM (expectation maximization) algorithm handles latent variables, while GMM is the Gaussian mixture model.