ABSTRACT
The coordination of many components of the movement system involved with most sports actions makes technique analysis a difficult endeavour. Usually discrete biomechanical variables are created in order to simplify the process. However, with converging acceptance among many researchers that coordination emerges by means of a self-organising interaction with all of the relevant constraints, the importance of an analysis tool for making these complex interactions discoverable becomes apparent. Artificial neural networks are able, through an iterative training process, to learn complex patterns, which make them attractive for analysing sports techniques. This chapter introduces self-organising maps (SOMs), a specific type of artificial neural network (ANN), as a potential tool for technique analysts, mainly because of the ability to provide a simple visualisation of the SOM output which represents the original complex movement pattern. A simple example to demonstrate SOMs is provided, followed by an overview of several techniques for visualising the output layer. Finally, a review of recent literature using SOMs for technique analysis is presented. Our outlook on the use of SOMs in technique analysis involves practitioners using them to focus on the coordination pattern itself, rather than narrowing the scope of the analysis to a few predefined key events in the action or in fact simply by judging the performance outcome.
