A Riemannian Newton trust-region method for fitting Gaussian mixture models

Abstract Gaussian Mixture Models are a powerful tool in Data Science and Statistics that mainly used for clustering density approximation. The task of estimating the model parameters is practice often solved by expectation maximization (EM) algorithm which has its benefits simplicity low per-iteration costs. However, EM converges slowly if there large share hidden information or overlapping clusters. Recent advances Manifold Optimization have gained increasing interest. We introduce an explicit formula Riemannian Hessian Models. On top, we propose new Newton Trust-Region method outperforms current approaches both terms runtime number iterations. apply our on problems approximation tasks. Our very data with compared to existing methods.