A Study in Deploying Self-Organized Map (SOM) in an Open Source J2EE Cluster and Caching System

Neural networks provide significant benefits in medical research. They are actively being used for such applications as locating previously undetected patterns in mountains of research data, controlling medical devices based on biofeedback, and detecting characteristics in medical imaging [Guan, H., et al., 1997]. Grouping of medical data based on key characteristics is a type of clustering problem. Neural networks can be used to solve clustering problems, typically through self-organized map (SOM) type network [Kohonen, T., 2001]. self-organized map (SOM) is a neural network algorithm used to represent and interpret large high-dimensional data sets in much lower dimensional spaces. It is invented by Professor Teuvo Kohonen, and is also known as Kohonen map. Even though SOM has been widely used in data analysis, the time required to train the map is high and therefore limits its usage. Recent years, different approaches have been conducted to tackle this problem [Roussinov, D., et al., 1998] and one is through the distributed computing technology. In this paper, we propose a model for developing and deploying a Self-Organized Map in an open source cluster and caching system under a popular distributed framework, J2EE. The objective of the study is to provide an efficient, flexible and low-cost model for implementing the SOM in a cluster environment. With the help of the cluster, any extra calculation work load required for a larger SOM can be accommodated easily by just adding more nodes or computers to the cluster.