Research on human activity recognition based on active learning

This paper addresses the problem of human activity recognition based on wearable sensors. In resent years researches on human daily activity recognition have enabled impressive result on substantial amount of labeled training samples. However, unlabeled samples are readily available but labeled ones are often difficult and slow to obtain. In order to reduce the level of supervision, this paper analyzes the feasibility of active learning for searching most informative samples to be labeled by a user in activity recognition. The Experimental results of daily human activity recognition indicate that the active learning approach can extract low-level context information from few sensor nodes and then be processed to obtain high-level context information; and the query functions can detect the informative unlabeled activity sample to ask people to label, so as to learn from large amount of readily available unlabeled data.