Title: Fast Gaussian mixture model estimation using online EM algorithms
Authors: Hien Nguyen - La Trobe University (Australia) [presenting]
Abstract: The use of the Gaussian mixture model for model-based clustering and classification is ubiquitous in the modern analysis of multivariate data. Unfortunately, the estimation of Gaussian mixture models is often computationally burdensome, especially when the dimension of the data and the number of observations are large. Modern trends in optimisation theory have leaned towards the usage of stochastic algorithms for high-dimensional and big data optimisation problems. A number of interesting online algorithms for stochastic estimation of Gaussian mixture models have been made available. We implement some of these algorithms in R and compare the performance of such algorithms against some algorithms that are currently available and implemented in R.