A new algorithm based on sequential pattern mining for person identification in ubiquitous environments

Abstract

This paper presents an approach to person identification in ubiquitous environments. Our approach uses the sequential pattern mining principle to extract frequent patterns in data collected from the different sensors disseminated in the ubiquitous environment. In contrast with existing, intrusive, person identification algorithms that have been proposed in the literature, where the data is basically composed of audiovisual or image files recorded during experiments, our approach is fully non-intrusive and is based on event sequences collected from heterogeneous sensors. Our approach is divided into three main phases: (1) frequent pattern mining,(2) assignment of weights to extracted patterns, and (3) classification. Experiments using data collected in the Domus and Testbed smart homes demonstrate that our approach accurately identifies persons and improves classification results, outperforming two of the approaches reported in the literature.