@inproceedings{du-etal-2018-variational,
title = "Variational Autoregressive Decoder for Neural Response Generation",
author = "Du, Jiachen and
Li, Wenjie and
He, Yulan and
Xu, Ruifeng and
Bing, Lidong and
Wang, Xuan",
editor = "Riloff, Ellen and
Chiang, David and
Hockenmaier, Julia and
Tsujii, Jun{'}ichi",
booktitle = "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing",
month = oct # "-" # nov,
year = "2018",
address = "Brussels, Belgium",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/D18-1354",
doi = "10.18653/v1/D18-1354",
pages = "3154--3163",
abstract = "Combining the virtues of probability graphic models and neural networks, Conditional Variational Auto-encoder (CVAE) has shown promising performance in applications such as response generation. However, existing CVAE-based models often generate responses from a single latent variable which may not be sufficient to model high variability in responses. To solve this problem, we propose a novel model that sequentially introduces a series of latent variables to condition the generation of each word in the response sequence. In addition, the approximate posteriors of these latent variables are augmented with a backward Recurrent Neural Network (RNN), which allows the latent variables to capture long-term dependencies of future tokens in generation. To facilitate training, we supplement our model with an auxiliary objective that predicts the subsequent bag of words. Empirical experiments conducted on Opensubtitle and Reddit datasets show that the proposed model leads to significant improvement on both relevance and diversity over state-of-the-art baselines.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="du-etal-2018-variational">
<titleInfo>
<title>Variational Autoregressive Decoder for Neural Response Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiachen</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenjie</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yulan</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lidong</namePart>
<namePart type="family">Bing</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2018-oct-nov</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ellen</namePart>
<namePart type="family">Riloff</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Chiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Julia</namePart>
<namePart type="family">Hockenmaier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun’ichi</namePart>
<namePart type="family">Tsujii</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Brussels, Belgium</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Combining the virtues of probability graphic models and neural networks, Conditional Variational Auto-encoder (CVAE) has shown promising performance in applications such as response generation. However, existing CVAE-based models often generate responses from a single latent variable which may not be sufficient to model high variability in responses. To solve this problem, we propose a novel model that sequentially introduces a series of latent variables to condition the generation of each word in the response sequence. In addition, the approximate posteriors of these latent variables are augmented with a backward Recurrent Neural Network (RNN), which allows the latent variables to capture long-term dependencies of future tokens in generation. To facilitate training, we supplement our model with an auxiliary objective that predicts the subsequent bag of words. Empirical experiments conducted on Opensubtitle and Reddit datasets show that the proposed model leads to significant improvement on both relevance and diversity over state-of-the-art baselines.</abstract>
<identifier type="citekey">du-etal-2018-variational</identifier>
<identifier type="doi">10.18653/v1/D18-1354</identifier>
<location>
<url>https://aclanthology.org/D18-1354</url>
</location>
<part>
<date>2018-oct-nov</date>
<extent unit="page">
<start>3154</start>
<end>3163</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Variational Autoregressive Decoder for Neural Response Generation
%A Du, Jiachen
%A Li, Wenjie
%A He, Yulan
%A Xu, Ruifeng
%A Bing, Lidong
%A Wang, Xuan
%Y Riloff, Ellen
%Y Chiang, David
%Y Hockenmaier, Julia
%Y Tsujii, Jun’ichi
%S Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing
%D 2018
%8 oct nov
%I Association for Computational Linguistics
%C Brussels, Belgium
%F du-etal-2018-variational
%X Combining the virtues of probability graphic models and neural networks, Conditional Variational Auto-encoder (CVAE) has shown promising performance in applications such as response generation. However, existing CVAE-based models often generate responses from a single latent variable which may not be sufficient to model high variability in responses. To solve this problem, we propose a novel model that sequentially introduces a series of latent variables to condition the generation of each word in the response sequence. In addition, the approximate posteriors of these latent variables are augmented with a backward Recurrent Neural Network (RNN), which allows the latent variables to capture long-term dependencies of future tokens in generation. To facilitate training, we supplement our model with an auxiliary objective that predicts the subsequent bag of words. Empirical experiments conducted on Opensubtitle and Reddit datasets show that the proposed model leads to significant improvement on both relevance and diversity over state-of-the-art baselines.
%R 10.18653/v1/D18-1354
%U https://aclanthology.org/D18-1354
%U https://doi.org/10.18653/v1/D18-1354
%P 3154-3163
Markdown (Informal)
[Variational Autoregressive Decoder for Neural Response Generation](https://aclanthology.org/D18-1354) (Du et al., EMNLP 2018)
ACL