Faculty Candidate Presentation: Autism and Human-Robot Interaction

[Jerry]

(Please try to attend this talk. A good audience helps us recruit good faculty --- JW)

THE DEPARTMENT OF COMPUTER SCIENCE

Presents an Invited Talk by Faculty Candidate

Karla Conn Welch

Psychophysiological Analysis of Affective States for Children with Autism
During Human-Computer and Human-Robot Interaction

Abstract

Given the importance of affective (emotional) cues in human-machine interaction and their significant role in autism intervention practice, the work covered in this seminar marks an important step towards intelligent intervention systems that embody human-like functionality â€" affect recognition and adaptation. Two computer-based cognitive tasks were designed for eliciting target affective states. Machine-learning algorithms were used to map between the collected physiological data and the intensity level of the affective states, as labeled on subjective reports. A physiology-based affect-inference and adaptation framework endows the technology (e.g., computer or robot) with the capability of detecting the affective states of a child with autism and responding to them accordingly. Experimental results on real-time affect recognition and adaptation are presented based on a pilot study of 6 children with autism in a proof-of-concept experiment involving a robot-based basketball game. The results demonstrate that such an affect-sensitive adaptive system could hold promise for computer/robot-assisted autism intervention and, more generally, could improve human-computer/human-robot interactions.

Tuesday, March 31, 2009
11:00 â€" 12:00
EB 0015


Ms.Welch will meet informally with students after her presentation.

[raptor]

Thoughts:

Candidate seemed very familiar with subject material, and very confident in presentation. 
Seemed very interested in the work she was presented.
Wasn't much focus on how CS tied into research work.  Big detractor for me.
Speaker was very quite and wasn't very dynamic when speaking.  This could make for a dull lecture.

Scott