Previous work on Human Activity Recognition (HAR) in smart homes has been focussed on extracting features from sensor data based on statistical information of an individual activity. But only few works take into account how that individual activity relates to sensor data acquired from the whole set of activities acquired throughout the entire recording period.
In this talk we report results on recent efforts to apply a popular information retrieval statistic, Term Frequency–Inverse Document Frequency (TF-IDF), with the objective of improving the accuracy of classifiers used for HAR. We extend existing work in this area by investigating how ensemble-based methods could be adopted for this approach.
Last modified:
This page is maintained by the seminar list administrator.