video thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Dynamic Bayesian Networks for Multimodal Interaction

Published on 2007-02-2510515 Views

Dynamic Bayesian networks (DBNs) offer a natural upgrade path beyond classical hidden Markov models and become especially relevant when temporal data contains higher order structure, multiple modaliti

Related categories

Presentation

Dynamic Bayesian Networks for Multimodal Interaction00:00
Outline02:00
Introduction04:17
Bayesian Networks06:54
Bayes Nets to Junction Trees11:01
Junction Tree Algorithm12:35
Maximum Likelihood with EM15:58
Dynamic Bayes Nets19:52
Two-Person Interaction24:48
DBN: Hidden ARMA Model26:31
Hidden ARMA Features:29:34
Conditional EM for hidden ARMA29:48
Conditional EM31:19
Hidden ARMA on Gesture33:51
DBN: Input-Output HMM34:11
Input-Output HMM Data35:04
Video Representation35:46
Input-Output HMM36:18
Input-Output HMM with CEM36:36
Input-Output HMM Results37:35
Intractable Dynamic Bayes Nets38:12
Intractable DBNs: Generalized EM38:43
Intractable DBNs Variational EM39:51
Dynamical System Trees40:44
DSTs and Generalized EM41:55
DSTs for American Football42:07
DSTs for Gene Networks43:19
Robotic Surgery, Haptics & Video44:26
Robotic Surgical Drills Results45:03
Conclusion46:26