Tuesday, August 26, 2008

MIT trains computers to organize data like humans

By Mass High Tech Staff

MIT researchers have programmed computers to think more like humans, according to the school.

Researchers have developed an algorithm to help computers recognize patterns in large sets of data, in a manner similar to humans. The model can analyze a set of data and figure out which type of organizational structure best fits it, according to MIT. The model considers a range of possible data structures, including trees, linear orders, rings, dominance hierarchies and clusters. It finds the best-fitting structure of each type for a given data set and then picks the type of structure that best represents the data.

Humans do the same thing, often unconsciously, according to the school. The development of the periodic table of the chemical elements and the organization of biological species into a tree-structured system of classification, are example of humans performing this kind of classification, the school said. Children exhibit this data organization skill when they learn that social networks can be organized into cliques, and that words like “dog” can fit into overlapping categories like “mammal” or “animal.”

The model could help scientists analyze large amounts of data and could also shed light on how the human brain discovers patterns.

Josh Tenenbaum, an MIT professor of brain and cognitive sciences, was the senor author on the paper published this month in the journal Proceedings of the National Academy of Science. The algorithm was developed by recent MIT Ph.D. recipient Charles Kemp, now an assistant professor of psychology at Carnegie Mellon University, along with Tenenbaum.

The research was funded by the James S. McDonnell Foundation Causal Learning Research Collaborative, the Air Force Office of Scientific Research, and the NTT Communication Sciences Laboratory.