@article{clark-etal-2014-locally,
title = "Locally Non-Linear Learning for Statistical Machine Translation via Discretization and Structured Regularization",
author = "Clark, Jonathan H. and
Dyer, Chris and
Lavie, Alon",
editor = "Lin, Dekang and
Collins, Michael and
Lee, Lillian",
journal = "Transactions of the Association for Computational Linguistics",
volume = "2",
year = "2014",
address = "Cambridge, MA",
publisher = "MIT Press",
url = "https://aclanthology.org/Q14-1031",
doi = "10.1162/tacl_a_00191",
pages = "393--404",
abstract = "Linear models, which support efficient learning and inference, are the workhorses of statistical machine translation; however, linear decision rules are less attractive from a modeling perspective. In this work, we introduce a technique for learning arbitrary, rule-local, non-linear feature transforms that improve model expressivity, but do not sacrifice the efficient inference and learning associated with linear models. To demonstrate the value of our technique, we discard the customary log transform of lexical probabilities and drop the phrasal translation probability in favor of raw counts. We observe that our algorithm learns a variation of a log transform that leads to better translation quality compared to the explicit log transform. We conclude that non-linear responses play an important role in SMT, an observation that we hope will inform the efforts of feature engineers.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="clark-etal-2014-locally">
<titleInfo>
<title>Locally Non-Linear Learning for Statistical Machine Translation via Discretization and Structured Regularization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jonathan</namePart>
<namePart type="given">H</namePart>
<namePart type="family">Clark</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chris</namePart>
<namePart type="family">Dyer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alon</namePart>
<namePart type="family">Lavie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2014</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<genre authority="bibutilsgt">journal article</genre>
<relatedItem type="host">
<titleInfo>
<title>Transactions of the Association for Computational Linguistics</title>
</titleInfo>
<originInfo>
<issuance>continuing</issuance>
<publisher>MIT Press</publisher>
<place>
<placeTerm type="text">Cambridge, MA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">periodical</genre>
<genre authority="bibutilsgt">academic journal</genre>
</relatedItem>
<abstract>Linear models, which support efficient learning and inference, are the workhorses of statistical machine translation; however, linear decision rules are less attractive from a modeling perspective. In this work, we introduce a technique for learning arbitrary, rule-local, non-linear feature transforms that improve model expressivity, but do not sacrifice the efficient inference and learning associated with linear models. To demonstrate the value of our technique, we discard the customary log transform of lexical probabilities and drop the phrasal translation probability in favor of raw counts. We observe that our algorithm learns a variation of a log transform that leads to better translation quality compared to the explicit log transform. We conclude that non-linear responses play an important role in SMT, an observation that we hope will inform the efforts of feature engineers.</abstract>
<identifier type="citekey">clark-etal-2014-locally</identifier>
<identifier type="doi">10.1162/tacl_a_00191</identifier>
<location>
<url>https://aclanthology.org/Q14-1031</url>
</location>
<part>
<date>2014</date>
<detail type="volume"><number>2</number></detail>
<extent unit="page">
<start>393</start>
<end>404</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Journal Article
%T Locally Non-Linear Learning for Statistical Machine Translation via Discretization and Structured Regularization
%A Clark, Jonathan H.
%A Dyer, Chris
%A Lavie, Alon
%J Transactions of the Association for Computational Linguistics
%D 2014
%V 2
%I MIT Press
%C Cambridge, MA
%F clark-etal-2014-locally
%X Linear models, which support efficient learning and inference, are the workhorses of statistical machine translation; however, linear decision rules are less attractive from a modeling perspective. In this work, we introduce a technique for learning arbitrary, rule-local, non-linear feature transforms that improve model expressivity, but do not sacrifice the efficient inference and learning associated with linear models. To demonstrate the value of our technique, we discard the customary log transform of lexical probabilities and drop the phrasal translation probability in favor of raw counts. We observe that our algorithm learns a variation of a log transform that leads to better translation quality compared to the explicit log transform. We conclude that non-linear responses play an important role in SMT, an observation that we hope will inform the efforts of feature engineers.
%R 10.1162/tacl_a_00191
%U https://aclanthology.org/Q14-1031
%U https://doi.org/10.1162/tacl_a_00191
%P 393-404
Markdown (Informal)
[Locally Non-Linear Learning for Statistical Machine Translation via Discretization and Structured Regularization](https://aclanthology.org/Q14-1031) (Clark et al., TACL 2014)
ACL