MIT EECS - Basis Technology Undergraduate Research and Innovation Scholar
Implementing a Bayesian Framework for Verb Learning from Syntax and Semantics
Robert C. Berwick
Natural language processing has achieved many impressive results, but the ways that computers learn language are very different from (and, so far, worse than) the ways that humans learn language. As a step towards improving the language acquisition model of computers, this project is to implement a system that independently learns the meanings of verbs, in a plausibly human-like way. The system will use both syntactic and semantic contextual information, modeled as features of verb meaning, to probabilistically converge on the meaning of a verb from only a few examples.
I worked at MIT Lincoln Laboratory on designing an experiment about connections between language and motor control in the brain. At the MIT Media Lab, I worked on implementing a visualization of the data flow in a cognitive model. I went on a volunteer service trip to Israel.