Classtering: A Semi-Supervised Algorithm Based on Mixture of Factor Analysers

In this work we propose a novel parametric Bayesian model to solve the problem of semi-supervised learning, including classification and clustering. Standard ap- proaches of semi-supervised classification can recognize classes but cannot find groups of data. On the other hand, semi-supervised clustering techniques are able to discover groups of data but cannot find the associations between clusters and classes. The proposed model can classify and cluster samples simultaneously, leading to the possibility of solving the problem of annotation in the presence of an unknown number of classes and/or arbitrary number of clusters per class. Preliminary results performed on synthetic datasets show the effectiveness of the framework.