Conditional Random Fields versus Hidden Markov Models for activity recognition in temporal sensor data

Open Access
Authors
Publication date 2008
Host editors
  • G.J.M. Smit
  • D.H.J. Epema
  • M.S. Lew
Book title Proceedings of the 14th Annual Conference of the Advanced School for Computing and Imaging
ISBN
  • 9789081084932
Event 14th Annual Conference of the Advanced School for Computing and Imaging (ASCI 2008), Heijen, the Netherlands
Publisher Delft: Advanced School for Computing and Imaging (ASCI)
Organisations
  • Faculty of Science (FNWI) - Informatics Institute (IVI)
Abstract Conditional Random Fields are a discriminative probabilistic model which recently gained popularity in applications that require modeling nonindependent observation sequences. In this work, we present the basic advantages of this model over generative models and argue about its suitability in the domain of activity recognition from sensor networks. We present experimental results on a realworld dataset that support this argumentation.
Document type Conference contribution
Published at http://www.science.uva.nl/research/isla/pub/kasteren08asci.pdf
Downloads
Permalink to this page
Back