Context Inference: Light-Weight Online Unsupervised Posture Detection by Smartphone Accelerometer

Abstract

This paper proposes a mild-weight online classification method to locate the user-centric postural movements, inclusive of sitting, standing, walking, and jogging, through smartphones. These moves are named user states considering that they're inferred after the evaluation of records acquired thru the accelerometer sensor built-in smartphones. To distinguish one-person country from some other, much research may be observed in the literature. But, this has a look at differs from others via supplying a computational mild-weight and online class method without understanding any a priori statistics. The proposed approach not only provides a standalone solution in the differentiation of user states but also assists other widely used class techniques by way of producing training records classes and/or enter device matrices. Further, this chapter intends to improve those present techniques for online processing. Eventually, the proposed approach nonetheless makes a solid differentiation in user states even wherein the sensor is being operated beneath slower sampling frequencies