BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Can Robots Learn Language the Way Children Do? - Stephen E. Levins
 on (University of Illinois)
DTSTART:20070710T103000Z
DTEND:20070710T113000Z
UID:TALK7664@talks.cam.ac.uk
CONTACT:Dr Marcus Tomalin
DESCRIPTION:Speech recognition machines are in use in more and more device
 s and\nservices.  Airlines\, banks\, and telephone companies provide\ninfo
 rmation to customers via spoken queries.  You can buy hand-held\ndevices\,
  appliances\, and PCs that are operated by spoken commands.\nAnd\, for aro
 und $100\, you can buy a program for your laptop that will\ntranscribe spe
 ech into text.  Unfortunately\, automatic speech\nrecognition systems are 
 quite error prone\, nor do they understand the\nmeanings of spoken message
 s in any significant way.  I argue that to\ndo so\, speech recognition mac
 hines would have to  possess the same\nkinds of cognitive abilities that h
 umans display.  Engineers have\nbeen trying to build machines with human-l
 ike abilities to think\nand use language for nearly 60 years without much 
 success.  Are all\nsuch efforts doomed to failure?  Maybe not.  I suggest\
 nthat if we take a radically different approach\, we might succeed.\nIf\, 
 instead of trying to program machines to behave intelligently\,\nwe design
  them to learn by experiencing the real world in the same\nway a child doe
 s\, we might solve the speech recognition problem\nin the process.  This i
 s the ambitious goal of the research now being\nconducted in my laboratory
 .  To date\, we have constructed three\nrobots that have attained some rud
 imentary visual navigation and\nobject manipulation abilities which they c
 an perform under spoken\ncommand.
LOCATION:LR4\, Engineering Department\, Baker Building
END:VEVENT
END:VCALENDAR