An essay in classifying Self-organizing Maps for temporal sequence processing

This paper presents a possible classification of modifications and adaptations of Self-organizing Maps (SOMs) for temporal sequence processing. Four main application areas for SOMs and temporal sequences have been identified. These are prediction, control, monitoring and mining. In order to model temporal relations among the data items within these application domains, usually an adaptation of the original learning algorithm, a modification of the network topology, or a combination of SOMs with special visualization techniques is made. Distinct approaches of SOMs for temporal sequence processing are classified into this scheme. Often, and in order to handle more complex domains, several adaptation forms are combined.