Distributed latent variable models of lexical co-occurrences

John Blitzer, Amir Globerson, Fernando Pereira

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Low-dimensional representations for lexical co-occurrence data have become increasingly important in alleviating the sparse data problem inherent in natural language processing tasks. This work presents a distributed latent variable model for inducing these low-dimensional representations. The model takes inspiration from both connectionist language models [1, 16] and latent variable models [13, 9]. We give results which show that the new model significantly improves both bigram and trigram models.

Original languageEnglish
Title of host publicationAISTATS 2005 - Proceedings of the 10th International Workshop on Artificial Intelligence and Statistics
Pages25-32
Number of pages8
StatePublished - 2005
Externally publishedYes
Event10th International Workshop on Artificial Intelligence and Statistics, AISTATS 2005 - Hastings, Christ Church, Barbados
Duration: 6 Jan 20058 Jan 2005

Publication series

NameAISTATS 2005 - Proceedings of the 10th International Workshop on Artificial Intelligence and Statistics

Conference

Conference10th International Workshop on Artificial Intelligence and Statistics, AISTATS 2005
Country/TerritoryBarbados
CityHastings, Christ Church
Period6/01/058/01/05

Fingerprint

Dive into the research topics of 'Distributed latent variable models of lexical co-occurrences'. Together they form a unique fingerprint.

Cite this