Extraction of attributes and knowledge rules for sport skill by TAM network

In this paper, we discuss sport technique evaluation of motion analysis modeled by TAM network as a kind of neural networks. We recorded continuous forehand strokes of each table tennis player into video frames, and analyzed the trajectory pattern of nine measurement markers attached at the body of players with the motion analysis model. We extracted input attributes and technique rules in order to classify the skill level of players of table tennis, i.e., expert player, middle level player and beginner. In addition, we analyzed movement of the markers in order to understand how to improve skill in table tennis technique.