A new soft assignment K-means algorithm

Abstract

K-means is one of the most popular and simple clustering algorithm. In spite of the fact that K-means was proposed over 60 years ago, it is still widely used. This paper provides a soft assignment K-means algorithm which is an extension of K-means where each data point can be a member of multiple clusters with a membership value. As an example, this paper apply soft assignment K-means algorithm to estimate the parameters of Gaussian mixture models and compare it with traditional K-means algorithm. Experiments demonstrate that soft assignment K-means algorithm can give more accurate result than traditional K-means algorithm which using hard assignment mechanism.