A batch self-organizing maps algorithm based on adaptive distances

Abstract

Clustering methods aims to organize a set of items into clusters such that items within a given cluster have a high degree of similarity, while items belonging to different clusters have a high degree of dissimilarity. The self-organizing map (SOM) introduced by Kohonen is an unsupervised competitive learning neural network method which has both clustering and visualization properties, using a neighborhood lateral interaction function to discover the topological structure hidden in the data set. In this paper, we introduce a batch self-organizing map algorithm based on adaptive distances. Experimental results obtained in real benchmark datasets show the effectiveness of our approach in comparison with traditional batch selforganizing map algorithms.

Symbolic Data Analysis: another look at the interaction of Data Mining and Statistics

“…Having a common representation setup based on a quantile‐vector representation (for a pre‐chosen set of quantiles) allows for a unified analysis of the data set by taking simultaneously into account variables of different types. Self‐Organizing Maps methodologies have been developed by Bock,; in Ref the authors introduce a batch self‐organizing map algorithm based on adaptive distances; in Ref an adaptive batch SOM method for Multiple Dissimilarity Data Tables is proposed; other approaches are investigated by Hajjar and Hamdan (see e.g., Refs ) and Yang et al Recently, a model‐based approach for clustering interval data has been developed, extending the Gaussian models proposed in Ref to the model‐based clustering context. For this purpose, the EM algorithm has been adapted for different covariance configurations.…”