Classifying tree structures using elastic matching of sequence encodings

Abstract

This document is the Accepted Manuscript version of the following article: Angeliki Skoura, Iosif Mporas, Vasileios Megalooikonomou, ???Classifying tree structures using elastic matching of sequence encodings???, Neurocomputing, Vol. 163, pp. 151-159, February 2015. The Version of Record is available online at: DOI: https://doi.org/10.1016/j.neucom.2014.08.083. This Manuscript version is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/ ), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.Structures of tree topology are frequently encountered in nature and in a range of scientific domains. In this paper, a multi-step framework is presented to classify tree topologies introducing the idea of elastic matching of their sequence encodings. Initially, representative sequences of the branching topologies are obtained using node labeling and tree traversal schemes. The similarity between tree topologies is then quantified by applying elastic matching techniques. The resulting sequence alignment reveals corresponding node groups providing a better understanding of matching tree topologies. The new similarity approach is explored using various classification algorithms and is applied to a medical dataset outperforming state-of-the-art techniques by at least 6.6% and 3.5% in terms of absolute specificity and accuracy correspondingly