Boosted learning in dynamic bayesian networks for multimodal speaker detection

Ashutosh Garg, Vladimir Pavlović, James M. Rehg

Research output: Contribution to journalArticlepeer-review

38 Scopus citations

Abstract

Bayesian network models provide an attractive framework for multitnodal sensor fusion. They combine an intuitive graphical representation with efficient algorithms for inference und learning. However, the unsupervised nature of standard parameter (earning algorithms for Bayesian networks can lead to poor performance in classification tasks. We have developed a supervised learning framework for Bayesian networks, which is based on the Adaboost algorithm of Schapire and Freund. Our framework covers static and dynamic Bayesian networks with both discrete and continuous states. We have tested our framework in the context of a novel multimodal HCI application: a speech-based command and control interface for a Smart Kiosk. We provide experimental evidence for the utility of our boosted learning approach.

Original languageEnglish (US)
Pages (from-to)1355-1369
Number of pages15
JournalProceedings of the IEEE
Volume91
Issue number9
DOIs
StatePublished - Sep 2003

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Keywords

  • Bayesian networks
  • Boosting
  • Discriminative learning
  • Muitimodal integration
  • Speaker detection

Fingerprint Dive into the research topics of 'Boosted learning in dynamic bayesian networks for multimodal speaker detection'. Together they form a unique fingerprint.

Cite this