@inproceedings{maupome-meurs-2020-language,
title = "Language Modeling with a General Second-Order {RNN}",
author = "Maupom{\'e}, Diego and
Meurs, Marie-Jean",
booktitle = "Proceedings of the 12th Language Resources and Evaluation Conference",
month = may,
year = "2020",
address = "Marseille, France",
publisher = "European Language Resources Association",
url = "https://aclanthology.org/2020.lrec-1.584",
pages = "4749--4753",
abstract = "Different Recurrent Neural Network (RNN) architectures update their state in different manners as the input sequence is processed. RNNs including a multiplicative interaction between their current state and the current input, second-order ones, show promising performance in language modeling. In this paper, we introduce a second-order RNNs that generalizes existing ones. Evaluating on the Penn Treebank dataset, we analyze how its different components affect its performance in character-lever recurrent language modeling. We perform our experiments controlling the parameter counts of models. We find that removing the first-order terms does not hinder performance. We perform further experiments comparing the effects of the relative size of the state space and the multiplicative interaction space on performance. Our expectation was that a larger states would benefit language models built on longer documents, and larger multiplicative interaction states would benefit ones built on larger input spaces. However, our results suggest that this is not the case and the optimal relative size is the same for both document tokenizations used.",
language = "English",
ISBN = "979-10-95546-34-4",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="maupome-meurs-2020-language">
<titleInfo>
<title>Language Modeling with a General Second-Order RNN</title>
</titleInfo>
<name type="personal">
<namePart type="given">Diego</namePart>
<namePart type="family">Maupomé</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marie-Jean</namePart>
<namePart type="family">Meurs</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2020-may</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<language>
<languageTerm type="text">English</languageTerm>
<languageTerm type="code" authority="iso639-2b">eng</languageTerm>
</language>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 12th Language Resources and Evaluation Conference</title>
</titleInfo>
<originInfo>
<publisher>European Language Resources Association</publisher>
<place>
<placeTerm type="text">Marseille, France</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-10-95546-34-4</identifier>
</relatedItem>
<abstract>Different Recurrent Neural Network (RNN) architectures update their state in different manners as the input sequence is processed. RNNs including a multiplicative interaction between their current state and the current input, second-order ones, show promising performance in language modeling. In this paper, we introduce a second-order RNNs that generalizes existing ones. Evaluating on the Penn Treebank dataset, we analyze how its different components affect its performance in character-lever recurrent language modeling. We perform our experiments controlling the parameter counts of models. We find that removing the first-order terms does not hinder performance. We perform further experiments comparing the effects of the relative size of the state space and the multiplicative interaction space on performance. Our expectation was that a larger states would benefit language models built on longer documents, and larger multiplicative interaction states would benefit ones built on larger input spaces. However, our results suggest that this is not the case and the optimal relative size is the same for both document tokenizations used.</abstract>
<identifier type="citekey">maupome-meurs-2020-language</identifier>
<location>
<url>https://aclanthology.org/2020.lrec-1.584</url>
</location>
<part>
<date>2020-may</date>
<extent unit="page">
<start>4749</start>
<end>4753</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Language Modeling with a General Second-Order RNN
%A Maupomé, Diego
%A Meurs, Marie-Jean
%S Proceedings of the 12th Language Resources and Evaluation Conference
%D 2020
%8 may
%I European Language Resources Association
%C Marseille, France
%@ 979-10-95546-34-4
%G English
%F maupome-meurs-2020-language
%X Different Recurrent Neural Network (RNN) architectures update their state in different manners as the input sequence is processed. RNNs including a multiplicative interaction between their current state and the current input, second-order ones, show promising performance in language modeling. In this paper, we introduce a second-order RNNs that generalizes existing ones. Evaluating on the Penn Treebank dataset, we analyze how its different components affect its performance in character-lever recurrent language modeling. We perform our experiments controlling the parameter counts of models. We find that removing the first-order terms does not hinder performance. We perform further experiments comparing the effects of the relative size of the state space and the multiplicative interaction space on performance. Our expectation was that a larger states would benefit language models built on longer documents, and larger multiplicative interaction states would benefit ones built on larger input spaces. However, our results suggest that this is not the case and the optimal relative size is the same for both document tokenizations used.
%U https://aclanthology.org/2020.lrec-1.584
%P 4749-4753
Markdown (Informal)
[Language Modeling with a General Second-Order RNN](https://aclanthology.org/2020.lrec-1.584) (Maupomé & Meurs, LREC 2020)
ACL