Investigation of self-organising dynamic neural tree networks

Abstract

Self Organising Dynamic Neural Tree Networks (DNTNs) provide hierarchical clustering that is potentially applicable to large data sets. Two DNTN models have been produced by Racz and Klotz and Li et al. Four DNTN variants have been developed and analysed in order to establish which of the four potential cluster expansion methods is the most robust to parameter alterations in the production of representative tree structures.