University of Minnesota
University Relations
http://www.umn.edu/urelate
612-624-6868
myU OneStop


Go to unit's home.

Home | Seminars and Symposia | Past seminars/symposia: Friday, October 29, 2004

DTC Seminar Series

The ESP Game, CAPTCHA, and Some Other Stuff: Utilizing the Power of Human Cycles

by

Luis von Ahn
Computer Science and Engineering
Carnegie Mellon University

Friday, October 29, 2004
3:30 pm

402 Walter Library

Luis von Ahn

We present several novel techniques for utilizing the computational abilities (or "cycles") of humans. Our techniques have a vast number of applications in areas as diverse as security, computer vision, accessibility of the Web, adult-content filtering, prevention of spam, and search. One family of techniques is CAPTCHA (http://www.captcha.net), automated tests that humans can pass but that current computer programs cannot. CAPTCHAs take advantage of the power of human cycles in order to differentiate people from computers, and have many applications in practical security. The previous results of our work, for instance, are used by Yahoo! to ensure that only humans obtain free email accounts.

Ahn presenting

Another family of techniques is exemplified by our work on the ESP Game (http://www.espgame.org). ESP is a seductive online game — many people play over 40 hours a week! — and when people play they help determine the content of images on the Web by providing meaningful labels for them (e.g., an image of a house gets labeled as "house"). If the game is played as much as other popular online games, all images on the Web can be labeled in just a few weeks. Attaching proper labels to all images on the Web would allow for more accurate image search engines, would improve the accessibility of Web sites (by providing descriptions of images to visually impaired individuals), and would help Web browsers block pornography. This approach to labeling images is simple but novel: rather than using computer vision techniques that don't work well enough, we encourage people to do the work for us by taking advantage of their desire to be entertained. In a few months, the ESP Game has collected over 4 million labels for images on the Web. This collection of manually labeled random images from the Web — probably the largest in the world — can already be used as training data to improve computer vision algorithms. Our initial work on both CAPTCHA and the ESP Game has already been covered by the major news outlets: CNN, The BBC, The New York Times, The Associated Press, USA Today, etc. Parts of this talk are joint work with Manuel Blum, Laura Dabbish, Nick Hopper and John Langford.

 

Luis von Ahn (BigLou@cs.cmu.edu) is currently a PhD student in the Computer Science Department at Carnegie Mellon University working with Professor Manuel Blum. His interests include ways to get people to work for him for free, artificial intelligence, theoretical cryptography, practical security, and human-computer interaction. He is the recipient of a Microsoft Research Fellowship.