@inproceedings{agrawal-etal-2025-multilingual,
title = "Multilingual Continual Learning using Attention Distillation",
author = "Agrawal, Sanjay and
Nayak, Deep and
Sembium, Vivek Varadarajan",
editor = "Rambow, Owen and
Wanner, Leo and
Apidianaki, Marianna and
Al-Khalifa, Hend and
Eugenio, Barbara Di and
Schockaert, Steven and
Darwish, Kareem and
Agarwal, Apoorv",
booktitle = "Proceedings of the 31st International Conference on Computational Linguistics: Industry Track",
month = jan,
year = "2025",
address = "Abu Dhabi, UAE",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.coling-industry.8/",
pages = "91--99",
abstract = "Query-product relevance classification is crucial for e-commerce stores like Amazon, ensuring accurate search results that match customer intent. Using a unified multilingual model across multiple languages/marketplaces tends to yield superior outcomes but also presents challenges, especially in maintaining performance across all languages when the model is updated or expanded to include a new one. To tackle this, we examine a multilingual continual learning (CL) framework focused on relevance classification tasks and address the issue of catastrophic forgetting. We propose a novel continual learning approach called attention distillation, which sequentially adds adapters for each new language and incorporates a fusion layer above language-specific adapters. This fusion layer distills attention scores from the previously trained fusion layer, focusing on the older adapters. Additionally, translating a portion of the new language data into older ones supports backward knowledge transfer. Our method reduces trainable parameters by 80{\%}, enhancing computational efficiency and enabling frequent updates, while achieving a 1-3{\%} ROC-AUC improvement over single marketplace baselines and outperforming SOTA CL methods on proprietary and external datasets."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="agrawal-etal-2025-multilingual">
<titleInfo>
<title>Multilingual Continual Learning using Attention Distillation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sanjay</namePart>
<namePart type="family">Agrawal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Deep</namePart>
<namePart type="family">Nayak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="given">Varadarajan</namePart>
<namePart type="family">Sembium</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-01</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 31st International Conference on Computational Linguistics: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Owen</namePart>
<namePart type="family">Rambow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leo</namePart>
<namePart type="family">Wanner</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marianna</namePart>
<namePart type="family">Apidianaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hend</namePart>
<namePart type="family">Al-Khalifa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Barbara</namePart>
<namePart type="given">Di</namePart>
<namePart type="family">Eugenio</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Steven</namePart>
<namePart type="family">Schockaert</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kareem</namePart>
<namePart type="family">Darwish</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Apoorv</namePart>
<namePart type="family">Agarwal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, UAE</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Query-product relevance classification is crucial for e-commerce stores like Amazon, ensuring accurate search results that match customer intent. Using a unified multilingual model across multiple languages/marketplaces tends to yield superior outcomes but also presents challenges, especially in maintaining performance across all languages when the model is updated or expanded to include a new one. To tackle this, we examine a multilingual continual learning (CL) framework focused on relevance classification tasks and address the issue of catastrophic forgetting. We propose a novel continual learning approach called attention distillation, which sequentially adds adapters for each new language and incorporates a fusion layer above language-specific adapters. This fusion layer distills attention scores from the previously trained fusion layer, focusing on the older adapters. Additionally, translating a portion of the new language data into older ones supports backward knowledge transfer. Our method reduces trainable parameters by 80%, enhancing computational efficiency and enabling frequent updates, while achieving a 1-3% ROC-AUC improvement over single marketplace baselines and outperforming SOTA CL methods on proprietary and external datasets.</abstract>
<identifier type="citekey">agrawal-etal-2025-multilingual</identifier>
<location>
<url>https://aclanthology.org/2025.coling-industry.8/</url>
</location>
<part>
<date>2025-01</date>
<extent unit="page">
<start>91</start>
<end>99</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Multilingual Continual Learning using Attention Distillation
%A Agrawal, Sanjay
%A Nayak, Deep
%A Sembium, Vivek Varadarajan
%Y Rambow, Owen
%Y Wanner, Leo
%Y Apidianaki, Marianna
%Y Al-Khalifa, Hend
%Y Eugenio, Barbara Di
%Y Schockaert, Steven
%Y Darwish, Kareem
%Y Agarwal, Apoorv
%S Proceedings of the 31st International Conference on Computational Linguistics: Industry Track
%D 2025
%8 January
%I Association for Computational Linguistics
%C Abu Dhabi, UAE
%F agrawal-etal-2025-multilingual
%X Query-product relevance classification is crucial for e-commerce stores like Amazon, ensuring accurate search results that match customer intent. Using a unified multilingual model across multiple languages/marketplaces tends to yield superior outcomes but also presents challenges, especially in maintaining performance across all languages when the model is updated or expanded to include a new one. To tackle this, we examine a multilingual continual learning (CL) framework focused on relevance classification tasks and address the issue of catastrophic forgetting. We propose a novel continual learning approach called attention distillation, which sequentially adds adapters for each new language and incorporates a fusion layer above language-specific adapters. This fusion layer distills attention scores from the previously trained fusion layer, focusing on the older adapters. Additionally, translating a portion of the new language data into older ones supports backward knowledge transfer. Our method reduces trainable parameters by 80%, enhancing computational efficiency and enabling frequent updates, while achieving a 1-3% ROC-AUC improvement over single marketplace baselines and outperforming SOTA CL methods on proprietary and external datasets.
%U https://aclanthology.org/2025.coling-industry.8/
%P 91-99
Markdown (Informal)
[Multilingual Continual Learning using Attention Distillation](https://aclanthology.org/2025.coling-industry.8/) (Agrawal et al., COLING 2025)
ACL