Gaussian Mixture Model is one type of soft clustering where we calculate probability of each data point before assign them to the clusters. Here, one data point can be assigned to more than one clusters. And the probability can be calculated using Bayes theorem. I hope this model performs better than the typical k-means clustering in terms of sum of square. You may use hybrid classifier but accuracy should be measured and it also depends on the characteristics of the input dataset.

http://statweb.stanford.edu/~tibs/stat315a/LECTURES/em.pdf