ACM TechNews

MIT Model Helps Computers Sort Data More Like Humans

MIT News (08/25/08) Trafton, Anne

Researchers at the Massachusetts Institute of Technology (MIT) have developed an algorithm that will enable computers to find patterns in data similar to a human. The model analyzes a set of data, such as trees, linear orders, rings, dominance hierarchies, and clusters, and then determines the organizational structure that best fits the data. "Instead of looking for a particular kind of structure, we came up with a broader algorithm that is able to look for all of these structures and weigh them against each other," says MIT professor Josh Tenenbaum. Scientists in a wide range of fields could use the model to analyze large amounts of data. The model also provides new clues on how the human brain finds order in sets of information, and could advance research in artificial intelligence. Tenenbaum assisted recent MIT Ph.D. recipient Charles Kemp in developing the algorithm.

http://web.mit.edu/newsoffice/2008/brain-data-0825.html

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.