@inproceedings{sauder-etal-2020-best,
title = "Best Student Forcing: A Simple Training Mechanism in Adversarial Language Generation",
author = "Sauder, Jonathan and
Hu, Ting and
Che, Xiaoyin and
Mordido, Goncalo and
Yang, Haojin and
Meinel, Christoph",
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-1.576",
pages = "4680--4688",
abstract = "Language models trained with Maximum Likelihood Estimation (MLE) have been considered as a mainstream solution in Natural Language Generation (NLG) for years. Recently, various approaches with Generative Adversarial Nets (GANs) have also been proposed. While offering exciting new prospects, GANs in NLG by far are nevertheless reportedly suffering from training instability and mode collapse, and therefore outperformed by conventional MLE models. In this work, we propose techniques for improving GANs in NLG, namely Best Student Forcing (BSF), a novel yet simple adversarial training mechanism in which generated sequences of high quality are selected as temporary ground-truth to further train the generator. We also use an ensemble of discriminators to increase training stability and sample diversity. Evaluation shows that the combination of BSF and multiple discriminators consistently performs better than previous GAN approaches over various metrics, and outperforms a baseline MLE in terms of Fr ́ech ́et Distance, a recently proposed metric capturing both sample quality and diversity.",
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="sauder-etal-2020-best">
<titleInfo>
<title>Best Student Forcing: A Simple Training Mechanism in Adversarial Language Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jonathan</namePart>
<namePart type="family">Sauder</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ting</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyin</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Goncalo</namePart>
<namePart type="family">Mordido</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haojin</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christoph</namePart>
<namePart type="family">Meinel</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>Language models trained with Maximum Likelihood Estimation (MLE) have been considered as a mainstream solution in Natural Language Generation (NLG) for years. Recently, various approaches with Generative Adversarial Nets (GANs) have also been proposed. While offering exciting new prospects, GANs in NLG by far are nevertheless reportedly suffering from training instability and mode collapse, and therefore outperformed by conventional MLE models. In this work, we propose techniques for improving GANs in NLG, namely Best Student Forcing (BSF), a novel yet simple adversarial training mechanism in which generated sequences of high quality are selected as temporary ground-truth to further train the generator. We also use an ensemble of discriminators to increase training stability and sample diversity. Evaluation shows that the combination of BSF and multiple discriminators consistently performs better than previous GAN approaches over various metrics, and outperforms a baseline MLE in terms of Fr ́ech ́et Distance, a recently proposed metric capturing both sample quality and diversity.</abstract>
<identifier type="citekey">sauder-etal-2020-best</identifier>
<location>
<url>https://aclanthology.org/2020.lrec-1.576</url>
</location>
<part>
<date>2020-may</date>
<extent unit="page">
<start>4680</start>
<end>4688</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Best Student Forcing: A Simple Training Mechanism in Adversarial Language Generation
%A Sauder, Jonathan
%A Hu, Ting
%A Che, Xiaoyin
%A Mordido, Goncalo
%A Yang, Haojin
%A Meinel, Christoph
%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 sauder-etal-2020-best
%X Language models trained with Maximum Likelihood Estimation (MLE) have been considered as a mainstream solution in Natural Language Generation (NLG) for years. Recently, various approaches with Generative Adversarial Nets (GANs) have also been proposed. While offering exciting new prospects, GANs in NLG by far are nevertheless reportedly suffering from training instability and mode collapse, and therefore outperformed by conventional MLE models. In this work, we propose techniques for improving GANs in NLG, namely Best Student Forcing (BSF), a novel yet simple adversarial training mechanism in which generated sequences of high quality are selected as temporary ground-truth to further train the generator. We also use an ensemble of discriminators to increase training stability and sample diversity. Evaluation shows that the combination of BSF and multiple discriminators consistently performs better than previous GAN approaches over various metrics, and outperforms a baseline MLE in terms of Fr ́ech ́et Distance, a recently proposed metric capturing both sample quality and diversity.
%U https://aclanthology.org/2020.lrec-1.576
%P 4680-4688
Markdown (Informal)
[Best Student Forcing: A Simple Training Mechanism in Adversarial Language Generation](https://aclanthology.org/2020.lrec-1.576) (Sauder et al., LREC 2020)
ACL