Ensembles of Topology Preserving Maps

Introduction

In the case of projection methods based on unsupervised learning, the problem of instability makes it difficult to devise an easy way of combining their results into a final solution. This is mainly because it is a global type of analysis, which means that very little information can be obtained once it has been completed.

This chapter centres on different methods of enhancing the use of ensembles with another visualisation-centred unsupervised learning algorithms: the topology-preserving map family.