Towards an artificial training expert system for basketball

Low power wearable devices equipped with powerful embedded processors found their place in applications such as personalized health, wellness/sports/fitness, rehabilitation, personal entertainment, social communications and lifestyle computing. In sports, the use of these devices together with powerful machine learning algorithms have opened up opportunity to create so called artificial training expert systems. In this work, an example of such system is discussed with the application in basketball training. The designed system is capable of recognizing basketball training type automatically by sampling data from battery powered wireless wearable device equipped with motion sensors. Support Vector Machine (SVM) is used as classification algorithm. The algorithm is capable of achieving 99.5% accuracy on the examined dataset.