UTCS Colloquium/AI-Regina Barzilay/MIT CSAIL: "Embracing Language Diversity: Unsupervised Multilingual Learning?" ACES 2.402, Friday, November 20, 2009, 11:00 a.m.
There is a sign-up schedule for this talk at http://www.cs.utex
Type of Tal
k: UTCS Colloquium/AI
Speaker/Affiliation: Regina Barzilay/MIT CSAIL
Date/Time: Friday, November 20, 2009/ 11:00 a.m.
Host: Ray Mooney
Talk Title: "Embracing Language Diversity:
Unsupervised Multilingual Learning?"
the deep connection between human languages has fascinated scholars, and
driven many important discoveries in linguistics and anthropology. In this
talk, I will show that this connection can empower unsupervised methods fo
r language analysis. The key insight is that joint learning from several la
nguages reduces uncertainty about the linguistic structure of each individu
I will present multilingual generative unsupervised model
s for morphological segmentation, part-of-speech tagging, and parsing. In
all of these instances we model the multilingual data as arising through a
combination of language-independent and language-specific probabilistic pr
ocesses. This feature allows the model to identify and learn from recurring
cross-lingual patterns to improve prediction accuracy in each language. I
will also discuss ongoing work on unsupervised decoding of ancient Ugaritic
tablets using data from related Semitic languages.
This is joint work
with Benjamin Snyder, Tahira Naseem and Jacob Eisenstein.