A self organizing map-harmony search hybrid algorithm for clustering biological data

Abstract

Self Organizing Map (SOM) is a significant algorithmic methodology to visualize data spaces of larger dimensions. Accurate analysis of the input data requires a well-trained SOM. Many measures are there in practice to analyse the quality of the map. One of the most commonly used measure is Quantization Error. A trained SOM grid with minimum quantization error may not be topologically well preserved. The quality of Topology preservation is measured using Topographic Error. Choosing SOM dimension for the map training procedure is not straight forward as it may not guarantee a map with minimum of quantization and topographic errors. This paper proposes an SOM-Harmony Hybrid algorithm that will compute the optimal dimension of the SOM grid with minimum values of topographic and quantization errors.