Exploring the applicability of Reservoir methods for Classifying Punctual Sports Activities Using On-body Sensors

Hunt, D., Parry, D. and Schliebs, S.

This paper explores the use of a reservoir computing (RC) method, Echo State Networks (ESN) to classify inertial sensor motion data collected from sensors worn by horse riders into punctual activities of interest within a scripted movement environment. RC methods incorporate both temporal and spatial aspects within the model and therefore may have applicability classifying signals with the varying temporal signatures often seen across activity instances even when performed by the same subject. ESN\'s, as one of a number of RC methods has a potential advantage in this case of being able to directly incorporate the inertial data into the reservoir without the need to segment this data into sliding windows. This is part of a wider set of work to build a wearable coach for technique feedback for Equestrian sport. Our use of RC methods on inertial data to classify punctual human activities is novel.

Cite as: Hunt, D., Parry, D. and Schliebs, S. (2014). Exploring the applicability of Reservoir methods for Classifying Punctual Sports Activities Using On-body Sensors. In Proc. Thirty-Seventh Australasian Computer Science Conference (ACSC 2014) Auckland, New Zealand. CRPIT, 147. Thomas, B. and Parry, D. Eds., ACS. 67-73

(from crpit.com) (local if available)