@inproceedings{han-etal-2024-efficient,
title = "Efficient Dynamic Hard Negative Sampling for Dialogue Selection",
author = "Han, Janghoon and
Lee, Dongkyu and
Shin, Joongbo and
Bae, Hyunkyung and
Bang, Jeesoo and
Kim, Seonghwan and
Choi, Stanley Jungkyu and
Lee, Honglak",
editor = "Nouri, Elnaz and
Rastogi, Abhinav and
Spithourakis, Georgios and
Liu, Bing and
Chen, Yun-Nung and
Li, Yu and
Albalak, Alon and
Wakaki, Hiromi and
Papangelis, Alexandros",
booktitle = "Proceedings of the 6th Workshop on NLP for Conversational AI (NLP4ConvAI 2024)",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.nlp4convai-1.6",
pages = "89--100",
abstract = "Recent studies have demonstrated significant improvements in selection tasks, and a considerable portion of this success is attributed to incorporating informative negative samples during training. While traditional methods for constructing hard negatives provide meaningful supervision, they depend on static samples that do not evolve during training, leading to sub-optimal performance. Dynamic hard negative sampling addresses this limitation by continuously adapting to the model{'}s changing state throughout training. However, the high computational demands of this method restrict its applicability to certain model architectures. To overcome these challenges, we introduce an efficient dynamic hard negative sampling (EDHNS). EDHNS enhances efficiency by pre-filtering easily discriminable negatives, thereby reducing the number of candidates the model needs to compute during training. Additionally, it excludes question-candidate pairs where the model already exhibits high confidence from loss computations, further reducing training time. These approaches maintain learning quality while minimizing computation and streamlining the training process. Extensive experiments on DSTC9, DSTC10, Ubuntu, and E-commerce benchmarks demonstrate that EDHNS significantly outperforms baseline models, proving its effectiveness in dialogue selection tasks.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="han-etal-2024-efficient">
<titleInfo>
<title>Efficient Dynamic Hard Negative Sampling for Dialogue Selection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Janghoon</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dongkyu</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joongbo</namePart>
<namePart type="family">Shin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyunkyung</namePart>
<namePart type="family">Bae</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jeesoo</namePart>
<namePart type="family">Bang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seonghwan</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stanley</namePart>
<namePart type="given">Jungkyu</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Honglak</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 6th Workshop on NLP for Conversational AI (NLP4ConvAI 2024)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Elnaz</namePart>
<namePart type="family">Nouri</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Abhinav</namePart>
<namePart type="family">Rastogi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Georgios</namePart>
<namePart type="family">Spithourakis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun-Nung</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alon</namePart>
<namePart type="family">Albalak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hiromi</namePart>
<namePart type="family">Wakaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alexandros</namePart>
<namePart type="family">Papangelis</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Recent studies have demonstrated significant improvements in selection tasks, and a considerable portion of this success is attributed to incorporating informative negative samples during training. While traditional methods for constructing hard negatives provide meaningful supervision, they depend on static samples that do not evolve during training, leading to sub-optimal performance. Dynamic hard negative sampling addresses this limitation by continuously adapting to the model’s changing state throughout training. However, the high computational demands of this method restrict its applicability to certain model architectures. To overcome these challenges, we introduce an efficient dynamic hard negative sampling (EDHNS). EDHNS enhances efficiency by pre-filtering easily discriminable negatives, thereby reducing the number of candidates the model needs to compute during training. Additionally, it excludes question-candidate pairs where the model already exhibits high confidence from loss computations, further reducing training time. These approaches maintain learning quality while minimizing computation and streamlining the training process. Extensive experiments on DSTC9, DSTC10, Ubuntu, and E-commerce benchmarks demonstrate that EDHNS significantly outperforms baseline models, proving its effectiveness in dialogue selection tasks.</abstract>
<identifier type="citekey">han-etal-2024-efficient</identifier>
<location>
<url>https://aclanthology.org/2024.nlp4convai-1.6</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>89</start>
<end>100</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Efficient Dynamic Hard Negative Sampling for Dialogue Selection
%A Han, Janghoon
%A Lee, Dongkyu
%A Shin, Joongbo
%A Bae, Hyunkyung
%A Bang, Jeesoo
%A Kim, Seonghwan
%A Choi, Stanley Jungkyu
%A Lee, Honglak
%Y Nouri, Elnaz
%Y Rastogi, Abhinav
%Y Spithourakis, Georgios
%Y Liu, Bing
%Y Chen, Yun-Nung
%Y Li, Yu
%Y Albalak, Alon
%Y Wakaki, Hiromi
%Y Papangelis, Alexandros
%S Proceedings of the 6th Workshop on NLP for Conversational AI (NLP4ConvAI 2024)
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F han-etal-2024-efficient
%X Recent studies have demonstrated significant improvements in selection tasks, and a considerable portion of this success is attributed to incorporating informative negative samples during training. While traditional methods for constructing hard negatives provide meaningful supervision, they depend on static samples that do not evolve during training, leading to sub-optimal performance. Dynamic hard negative sampling addresses this limitation by continuously adapting to the model’s changing state throughout training. However, the high computational demands of this method restrict its applicability to certain model architectures. To overcome these challenges, we introduce an efficient dynamic hard negative sampling (EDHNS). EDHNS enhances efficiency by pre-filtering easily discriminable negatives, thereby reducing the number of candidates the model needs to compute during training. Additionally, it excludes question-candidate pairs where the model already exhibits high confidence from loss computations, further reducing training time. These approaches maintain learning quality while minimizing computation and streamlining the training process. Extensive experiments on DSTC9, DSTC10, Ubuntu, and E-commerce benchmarks demonstrate that EDHNS significantly outperforms baseline models, proving its effectiveness in dialogue selection tasks.
%U https://aclanthology.org/2024.nlp4convai-1.6
%P 89-100
Markdown (Informal)
[Efficient Dynamic Hard Negative Sampling for Dialogue Selection](https://aclanthology.org/2024.nlp4convai-1.6) (Han et al., NLP4ConvAI-WS 2024)
ACL
- Janghoon Han, Dongkyu Lee, Joongbo Shin, Hyunkyung Bae, Jeesoo Bang, Seonghwan Kim, Stanley Jungkyu Choi, and Honglak Lee. 2024. Efficient Dynamic Hard Negative Sampling for Dialogue Selection. In Proceedings of the 6th Workshop on NLP for Conversational AI (NLP4ConvAI 2024), pages 89–100, Bangkok, Thailand. Association for Computational Linguistics.