Colloquium Details
Computational Behavioral Science: A New Paradigm for Interdisciplinary Research
| Author: | Jan van Santen Center for Spoken Language Understanding, Oregon Health & Science University |
|---|---|
| Date: | April 26, 2012 |
| Time: | 10:30 |
| Location: | 220 Deschutes |
| Host: | College of Arts and Sciences |
Abstract
The difficulty of objectively measuring human speech, language, movement, facial expression, and gesture, is a serious obstacle for many endeavors, whether clinical diagnosis, phenotypic characterization for genetics studies, or educational assessment. However, over the past few decades there had been steady progress on multiple fronts in computational science that has now produced tools that in combination can be brought to bear on this challenge. These tools derive from a variety of computational areas, such as computer vision, signal processing, computational linguistics, pattern recognition, machine learning, and the availability of vast amounts of online data, to mention a few. I use the term “Computational Behavioral Science” to refer to research on new algorithms for measuring or influencing human behavior, with, in my case, a special focus on neurological disorders.
In this talk, I will discuss a couple of examples, including, time permitting: detection of autism using analysis of speech prosody and language use; talk-through device that enhances the intelligibility of dysarthric speech; and automated voice based cognitive assessment. My colleagues Roark and Shafran will highlight other examples of Computational Behavioral Science in their talks.
I will conclude by emphasizing the wide applicability of Computational Behavioral Science, not only to neurological disorders other than the ones discussed, but also to educational assessment, neuroscience research, and linguistics research. Last, but not least, the computational challenges posed by Computational Behavioral Science are considerable, and may very well become a new class of drivers for Computer Science.
Biography
Jan van Santen (Ph.D., Mathematical Psychology, Michigan, 1979), worked until 1984 on human visual perception and image processing at Bell Labs and at New York University, and from 1986 until presently on Spoken Language Technology (SLT), first again at Bell Labs until 2000 and then at the Oregon Graduate Institute (acquired by the Oregon Health & Science University in 2001), where he became the Director of the Center for Spoken Language Understanding. At Bell Labs, he focused on Text-to-Speech Synthesis, specifically quantitative modeling of prosody. Since moving to OGI, he has worked primarily on applying SLT methods to neurological disorders, specifically neurodevelopmental disorders such as autism and neurodegenerative disorders such as Parkinson’s. He has over 100 publications and 7 patents.