Sciweavers

Share
ICPR
2002
IEEE

Boosting and Structure Learning in Dynamic Bayesian Networks for Audio-Visual Speaker Detection

10 years 9 months ago
Boosting and Structure Learning in Dynamic Bayesian Networks for Audio-Visual Speaker Detection
Bayesian networks are an attractive modeling tool for human sensing, as they combine an intuitive graphical representation with ef?cient algorithms for inference and learning. Earlier work has demonstrated that boosted parameter learning could be used to improve the performance of Bayesian network classi?ers for complex multi-modal inference problems such as speaker detection. In speaker detection, the goal is to use video and audio cues to infer when a person is speaking to a user interface. In this paper we introduce a new boosted structure learning algorithm based on AdaBoost. Given labeled data, our algorithm modi?es both the network structure and parameters so as to improve classi?cation accuracy. We compare its performance to both standard structure learning and boosted parameter learning on a ?xed structure. We present results for speaker detection and for the UCI "chess" dataset.
Tanzeem Choudhury, James M. Rehg, Vladimir Pavlovi
Added 09 Nov 2009
Updated 09 Nov 2009
Type Conference
Year 2002
Where ICPR
Authors Tanzeem Choudhury, James M. Rehg, Vladimir Pavlovic, Alex Pentland
Comments (0)
books