Template Trees
Matthew Partridge, M. Jabri, · Journal of Intelligent Systems · 1997
There are many applications for automated classifiers, including tasks such as diagnosis, recognition and identification.Decision trees have had many successes as classifiers.This article introduces a novel template based decision tree, or template tree.A template tree is a multivariate decision tree which makes a binary decision at each node by comparing an unknown instance with the template at each node.It has a hierarchical structure which means that classification can be extremely fast.However, because it is based on template-matching, its classification process can be understood, and can be more appropriate on high dimensional tasks.This article discusses how such a template tree classifies, how one can be constructed and shows variations on the framework.Template trees offer significantly better performance on the gene database than all other tested techniques, interpretative classification and an impressive recognition time.