Research on Variational Bayesian-based Data Classification Algorithm

Abstract

With the rapid development of Internet technology, the size and complexity of the database are continually growing, the traditional classification method can no longer meet the demand of the classification of complex data. As for such problems, a data classification algorithm based on variational bayesian is proposed. The algorithm introduce variational approximation theory on the basis of traditional bayesian inference, combined with the thought of maximum expected algorithm, utilizing the mean field theory in the statistical physics, and take gaussian mixture model as an experiment simulation. The experimental results show that randomly generated data can be seen clearly mixtured by three groups of gaussian model after 382 iterations, lower bound of likelihood function rises with the increase of iteration number, curve becomes flat as expected after 350 iterations, and getting mean value and the inverse of covariance matrix close to the real data in the range of allowable error, realizing its classification processing. Under the requirement of high precision, its calculation speed is faster, calculation efficiency is higher, which can in accordance with actual engineering application background.