view MIT-media-projects.org @ 333:f4ef73370da1

add file.
author Robert McIntyre <rlm@mit.edu>
date Fri, 20 Jul 2012 11:21:04 -0500
parents
children c264ebf683b4
line wrap: on
line source
1 *Machine Learning and Pattern Recognition with Multiple Modalities
2 Hyungil Ahn and Rosalind W. Picard
4 This project develops new theory and algorithms to enable computers to
5 make rapid and accurate inferences from multiple modes of data, such
6 as determining a person's affective state from multiple sensors—video,
7 mouse behavior, chair pressure patterns, typed selections, or
8 physiology. Recent efforts focus on understanding the level of a
9 person's attention, useful for things such as determining when to
10 interrupt. Our approach is Bayesian: formulating probabilistic models
11 on the basis of domain knowledge and training data, and then
12 performing inference according to the rules of probability
13 theory. This type of sensor fusion work is especially challenging due
14 to problems of sensor channel drop-out, different kinds of noise in
15 different channels, dependence between channels, scarce and sometimes
16 inaccurate labels, and patterns to detect that are inherently
17 time-varying. We have constructed a variety of new algorithms for
18 solving these problems and demonstrated their performance gains over
19 other state-of-the-art methods.
21 http://affect.media.mit.edu/projectpages/multimodal/