EM algorithm based MDL application to estimate the mixture model clustering parameters

The paper presents a method of mixture model clustering for multidimensional data. A novel technique is presented in this paper in order to aid in an improved the clustering performance, which is called minimum description length (MDL). The technique attempts to find the model order which minimizes the number of bits that would be required to code both the data samples and the parameters vector. It also includes an unsupervised method for estimating the number of cluster and the parameters of the model sequentially which is called clustered components analysis (CCA). Lastly, our method is applied to simulated data for verification.