eduzhai > Applied Sciences > Engineering >

ESPRESSO Entropy and ShaPe awaRe timE-Series SegmentatiOn for processing heterogeneous sensor data

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 23 pages

Abstract: Extracting informative and meaningful temporal segments from high-dimensionalwearable sensor data, smart devices, or IoT data is a vital preprocessing stepin applications such as Human Activity Recognition (HAR), trajectoryprediction, gesture recognition, and lifelogging. In this paper, we proposeESPRESSO (Entropy and ShaPe awaRe timE-Series SegmentatiOn), a hybridsegmentation model for multi-dimensional time-series that is formulated toexploit the entropy and temporal shape properties of time-series. ESPRESSOdiffers from existing methods that focus upon particular statistical ortemporal properties of time-series exclusively. As part of model development, anovel temporal representation of time-series $WCAC$ was introduced along with agreedy search approach that estimate segments based upon the entropy metric.ESPRESSO was shown to offer superior performance to four state-of-the-artmethods across seven public datasets of wearable and wear-free sensing. Inaddition, we undertake a deeper investigation of these datasets to understandhow ESPRESSO and its constituent methods perform with respect to differentdataset characteristics. Finally, we provide two interesting case-studies toshow how applying ESPRESSO can assist in inferring daily activity routines andthe emotional state of humans.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×