Detecting label errors in crowd-sourced smartphone sensor data

Xiao Bo, Christian Poellabauer, Megan K. Obrien, Chaithanya Krishna Mummidisetty, Arun Jayaraman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

Applications relying on supervised learning algorithms are susceptible to producing false outputs in the presence of label errors, i.e., situations were labels have been corrupted, both deliberately and accidentally. While prior work has focused on detecting and handling label errors for various types of applications, there is a lack of research addressing label errors in smartphone-based crowd-sensing applications, especially when used for action recognition. In this paper, we discuss and address two common types of smartphone-based label errors:mislabeling and multi-action labels. We also compare multiple learning algorithms, including an ensemble of four stratified trained classifiers. The results indicate the importance of the action type for filtering label error. The goal of this work is to provide guidelines for developing effective techniques to discover and remove error labels for action recognition systems.

Original languageEnglish (US)
Title of host publicationProceedings - 3rd International Workshop on Social Sensing, SocialSens 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages20-25
Number of pages6
ISBN (Electronic)9781538661659
DOIs
StatePublished - May 25 2018
Event3rd International Workshop on Social Sensing, SocialSens 2018 - Orlando, United States
Duration: Apr 17 2018 → …

Publication series

NameProceedings - 3rd International Workshop on Social Sensing, SocialSens 2018

Other

Other3rd International Workshop on Social Sensing, SocialSens 2018
CountryUnited States
CityOrlando
Period4/17/18 → …

Keywords

  • Action Recognition
  • Crowd-Sourcing
  • Ensemble Learning
  • Label Error

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Media Technology
  • Social Psychology
  • Communication

Fingerprint Dive into the research topics of 'Detecting label errors in crowd-sourced smartphone sensor data'. Together they form a unique fingerprint.

  • Cite this

    Bo, X., Poellabauer, C., Obrien, M. K., Mummidisetty, C. K., & Jayaraman, A. (2018). Detecting label errors in crowd-sourced smartphone sensor data. In Proceedings - 3rd International Workshop on Social Sensing, SocialSens 2018 (pp. 20-25). (Proceedings - 3rd International Workshop on Social Sensing, SocialSens 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SocialSens.2018.00017