Affective Computing


Recently, momentum is increasing in the belief that emotion is a critical part of intelligence. Further, some believe that true intelligence may not be possible without also including emotion into the model. The Affective Computing group at MIT, under the direction of Rosalind Picard, has been at the forefront of research in this area. Rosalind's group focuses on a wide spectrum of research areas, from computers sensing affect in humans to synthesizing emotions in computers.

Sensing emotions in users is an important aspect of future user interfaces. For example, if a computer could sense that its user was getting frustrated, it could change the way in which information was being presented. A vehicle equipped with affect sensing equipment might be able to detect if a driver was falling asleep, or impaired due to alcohol or drugs. The group uses a variety of sensors to evaluate the user's affective state. For example, a galvanic skin response sensor (GSR) for measuring skin conductance can identify the user's anxiety or a startle response. A respiration sensor can measure the rate and depth of a user 's breathing , which can be used to identify the user's state of alertness. Finally, an electromyogram sensor can measure the electromygraphic activity of muscles , detecting muscle contraction. The affective computing group has successfully measure jaw clenching with this device that could be used to identify states of anger in the user.

Sensing the external effects of emotion is one task, but identifying what they represent for an individual user is another. The group also researches recognizing (and learning) emotive states based upon a set of externally visible markers. Since these can be different for each person, the recognition task must including learning so that physical measurements properly map to defined states of emotion.

The ability to synthesize emotions in intelligent machines is another interesting area of research within MIT's group. They focus on not only synthesizing emotions that are externally understood , but also modeling true emotions as an internal mechanism. Consider a spacecraft subsystem that modeled varying types of emotion. If the spacecraft had lost pointing to a ground station on the Earth, it could integrate emotions of fear and raised state of awareness about its position. Consider also that while in this state, its lack of pointing causes it to decrease its available store of power in its onboard batteries. In this scenario, the spacecraft's pointing becomes even more critical as complete loss of power can lead to the loss (death) of the spacecraft. The incremental fear " experienced " by the spacecraft would lead to progressively more drastic measures in order to gain control, leading to decreased degrees of fear and an increasing state of happiness. While the spacecraft is in a state of fear, it would concentrate less on tasks that don't affect the survival of the spacecraft (such as onboard science experiments) and more on those aspects that decrease the fear and lead to happiness. While this scenario is imaginative, affective states can offer new ways to think about fault management and recovery, in addition to dealing with combinations of failures that may or may not have similar mechanisms for recovery.




Visual Basic Developer
Visual Basic Developers Guide to ASP and IIS: Build Powerful Server-Side Web Applications with Visual Basic. (Visual Basic Developers Guides)
ISBN: 0782125573
EAN: 2147483647
Year: 1999
Pages: 175

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net