Visualization Architecture Based on SOM for Two-Class Sequential Data

Abstract

In this paper, we propose a visualization architecture that constructs a map suggesting clusters in sequence that involve classification utilizing the class label information for the display method of the map. This architecture is based on Self-Organizing Maps (SOM) that are to create clusters and to arrange the similar clusters near within the low dimensional map. This proposed method consists of three steps, firstly the winner neuron trajectories are obtained by SOM, secondly, connectivity weights are obtained by a single layer perceptron based on the winner neuron trajectories, finally, the map is visualized by reversing the obtained weights into the map. In the experiments using time series of real-world medical data, we evaluate the visualization and classification performance by comparing the display method by the number of sample ratio for classes belonging to each cluster.