Difference between revisions of "Berg-Kirkpatrick et al, ACL 2010: Painless Unsupervised Learning with Features"
From Cohen Courses
Jump to navigationJump to search (Created page with '== Citation == T. Berg-Kirkpatrick, A. Bouchard-Côté, J. DeNero, and D. Klein. '''Painless Unsupervised Learning with Features''', ''Human Language Technologies 2010'', pp. 58…') |
|||
| Line 1: | Line 1: | ||
| + | __TOC__ | ||
== Citation == | == Citation == | ||
| Line 6: | Line 7: | ||
[http://www.aclweb.org/anthology-new/N/N10/N10-1083.pdf Download] | [http://www.aclweb.org/anthology-new/N/N10/N10-1083.pdf Download] | ||
| + | |||
| + | == Summary == | ||
| + | |||
| + | This paper generalizes conventional HMMs to featurized HMMs, by replacing the multinomial conditional probability distributions (CPDs) with miniature log-linear models. Two algorithms for unsupervised training of featurized HMMs are proposed. | ||
| + | |||
| + | Featurized HMMs are applied to four unsupervised learning tasks: | ||
| + | * POS induction (unsupervised version of POS tagging); | ||
| + | * Grammar induction; | ||
| + | * Word alignment; | ||
| + | * Word segmentation. | ||
| + | For all these four tasks, featurized HMMs are shown to outperform their unfeaturized counterparts by a substantial margin. | ||
| + | |||
| + | == Featurized HMMs == | ||
| + | |||
| + | === Definition === | ||
| + | |||
| + | === The Estimation Problem === | ||
| + | |||
| + | === The Decoding Problem === | ||
| + | |||
| + | === The Training Problem === | ||
| + | |||
| + | == Experiments == | ||
| + | |||
| + | === POS Induction === | ||
| + | |||
| + | === Grammar Induction === | ||
| + | |||
| + | === Word Alignment === | ||
| + | |||
| + | === Word Segmentation === | ||
Revision as of 16:40, 17 September 2011
Contents
Citation
T. Berg-Kirkpatrick, A. Bouchard-Côté, J. DeNero, and D. Klein. Painless Unsupervised Learning with Features, Human Language Technologies 2010, pp. 582-590, Los Angeles, June 2010.
Online Version
Summary
This paper generalizes conventional HMMs to featurized HMMs, by replacing the multinomial conditional probability distributions (CPDs) with miniature log-linear models. Two algorithms for unsupervised training of featurized HMMs are proposed.
Featurized HMMs are applied to four unsupervised learning tasks:
- POS induction (unsupervised version of POS tagging);
- Grammar induction;
- Word alignment;
- Word segmentation.
For all these four tasks, featurized HMMs are shown to outperform their unfeaturized counterparts by a substantial margin.
