The EM Algorithm and its Packages in R Project: A Literature Review
Haibin Qiu, Yanan Song, Tingdi Zhao · International Conference on Electric Information and Control Engineering · 2012
Expectation-maximization algorithm is a broadly applicable algorithm in statistics which was formally proposed in 1977. It is widely used in medical image reconstruction, data clustering in machine learning and computer vision, and estimating item parameters and latent abilities of item response theory models in psychometrics, and so forth. It is an iterative method for ending maximum likelihood or maximum posteriori estimates of parameters in statistical models. Many researchers work for its improving, such as generalized expectation maximization (GEM) and expectation conditional maximization (ECM). EM algorithm can be implemented in R project and the using of R project in EM algorithm just emerged in recent years. In this paper, the description and definition of EM algorithm will be mentioned firstly. Then we describe the researches on the main drawbacks of the EM algorithm which are its slow convergence and the dependence of the solution on both the stopping criterion and the initial values used. A brief introduction about hap assoc-package, EM Jump Diffusion-pagckge and Turbo EM-package is given which is the implementation of EM algorithm in R project. It is concluded that developing a total and integrated R project package for EM algorithm is necessary and possible.