Research Projects NTT-MIT Research Collaboration: a partnership in the future of communication and computation

Theory and Application of Concept Learning


Start date: 07/2002

Joshua Tennenbaum

Naonori Ueda

Project summary

Our project has two threads:

Project description


Two specific questions drive this research program.

  • First, how do people -- and how can computers -- combine unsupervised and supervised approaches to concept learning, using unsupervised learning to build better hypothesis spaces for supervised concept learning?
  • Second, what is the large-scale structure of human semantic concepts in natural language, and how can this structure be learned – by either people or machines -- from large text or hypertext corpora?

Demos, movies and other examples

The principal investigators

Presentations and posters



Proposals and progress reports


Progress Reports:

NTT Bi-Annual Progress Report, July to December 2002:

For more information