Heuristic and optimization approaches to extending the Kohonen self organizing algorithm

The Kohonen self organizing neural network has been applied to an increasingly wider range of application problems that traditionally have been the domain of statistical and operational research techniques, such as data clustering and classification, and optimization and control. This Kohonen network is bestowed with a number of unique strengths which are, unfortunately, matched by an equally formidable set of limitations due to its learning algorithm. There have been extensive studies over the last decade to extend the Kohonen neural network using heuristic and optimization approaches. This paper provides a comprehensive survey of the research efforts directed to enhancing the Kohonen self organizing neural network and its learning algorithm. We also point out some research directions for pursuing further improvements.