Author: Zen and the Art of Computer Programming
1 Introduction
Gaussian mixture model (GMM) is a clustering method based on probability distribution that can be used to describe the distribution of data in high-dimensional data sets. This method assumes that each data point is a multivariate Gaussian model composed of a mixture of multiple Gaussian distributions, and uses the EM algorithm to iteratively optimize the model parameters to maximize the possibility that the data points belong to each Gaussian distribution. Therefore, GMM is an unsupervised learning method that can train clustering effects without label information. It is a typical unsupervised learning algorithm.
This article will provide a systematic, comprehensive, and easy-to-understand explanation of GMM, hoping to help readers quickly understand the working mechanism of GMM and master the application of GMM. At the same time, I also hope that readers can provide valuable opinions and give me a more complete tutorial.
2.Basic concepts and terminology of GMM
(1) Model definition
GMM consists of multivariate Gaussian distribution, that is, p ( x ∣ θ ) p(x\mid \theta)p</