Human activity recognition in big data smart home context

Abstract

In smart homes, a few millions of data are sent by different sensors and RFID per day. It thus constitutes a Big Data warehouse with all the problems that come from and in our work, we are interested in the problem of managing huge data in streaming. In this paper, we propose to use very fast decision tree(VFDT) for activity recognition. We formulate Activity Recognition as a classification problem where classes correspond to activities. The performance of the VFDT and other classifiers is compared.