Title :
Semi-supervised meeting event recognition with adapted HMMs
Author :
Zhang, Dong ; Gatica-Perez, Daniel ; Bengio, Samy
Author_Institution :
IDIAP Res. Inst., Swiss Fed. Inst. of Technol., Lausanne, Switzerland
Abstract :
This paper investigates the use of unlabeled data to help labeled data for audio-visual event recognition in meetings. To deal with situations in which it is difficult to collect enough labeled data to capture event characteristics, but collecting a large amount of unlabeled data is easy, we present a semi-supervised framework using HMM adaptation techniques. Instead of directly training one model for each event, we first train a well-estimated general event model for all events using both labeled and unlabeled data, and then adapt the general model to each specific event model using its own labeled data. We illustrate the proposed approach with a set of eight audio-visual events defined in meetings. Experiments and comparison with the fully-supervised baseline method show the validity of the proposed semi-supervised approach.
Keywords :
hidden Markov models; image recognition; learning (artificial intelligence); speech recognition; HMM adaptation technique; audio-visual event; meeting event recognition; semisupervised framework; Ambient intelligence; Character recognition; Concatenated codes; Event detection; Hidden Markov models; Labeling; Speech analysis; Surveillance; Testing; Training data;
Conference_Titel :
Multimedia and Expo, 2005. ICME 2005. IEEE International Conference on
Print_ISBN :
0-7803-9331-7
DOI :
10.1109/ICME.2005.1521618