@inproceedings{prasad-etal-2025-faster,
title = "Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction",
author = "Prasad, Kritarth and
Zaki, Mohammadi and
Singh, Pratik Rakesh and
Wasnik, Pankaj",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Findings of the Association for Computational Linguistics: NAACL 2025",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-naacl.466/",
doi = "10.18653/v1/2025.findings-naacl.466",
pages = "8322--8335",
ISBN = "979-8-89176-195-7",
abstract = "Ensembling neural machine translation (NMT) models to produce higher-quality translations than the $L$ individual models has been extensively studied. Recent methods typically employ a candidate selection block (CSB) and an encoder-decoder fusion block (FB), requiring inference across \textit{all} candidate models, leading to significant computational overhead, generally $\Omega(L)$. This paper introduces \textbf{SmartGen}, a reinforcement learning (RL)-based strategy that improves the CSB by selecting a small, fixed number of candidates and identifying optimal groups to pass to the fusion block for each input sentence. Furthermore, previously, the CSB and FB were trained independently, leading to suboptimal NMT performance. Our DQN-based \textbf{SmartGen} addresses this by using feedback from the FB block as a reward during training. We also resolve a key issue in earlier methods, where candidates were passed to the FB without modification, by introducing a Competitive Correction Block (CCB). Finally, we validate our approach with extensive experiments on English-Hindi translation tasks in both directions as well as English to Chinese and English to German."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="prasad-etal-2025-faster">
<titleInfo>
<title>Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kritarth</namePart>
<namePart type="family">Prasad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammadi</namePart>
<namePart type="family">Zaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pratik</namePart>
<namePart type="given">Rakesh</namePart>
<namePart type="family">Singh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pankaj</namePart>
<namePart type="family">Wasnik</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: NAACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-195-7</identifier>
</relatedItem>
<abstract>Ensembling neural machine translation (NMT) models to produce higher-quality translations than the L individual models has been extensively studied. Recent methods typically employ a candidate selection block (CSB) and an encoder-decoder fusion block (FB), requiring inference across all candidate models, leading to significant computational overhead, generally Ømega(L). This paper introduces SmartGen, a reinforcement learning (RL)-based strategy that improves the CSB by selecting a small, fixed number of candidates and identifying optimal groups to pass to the fusion block for each input sentence. Furthermore, previously, the CSB and FB were trained independently, leading to suboptimal NMT performance. Our DQN-based SmartGen addresses this by using feedback from the FB block as a reward during training. We also resolve a key issue in earlier methods, where candidates were passed to the FB without modification, by introducing a Competitive Correction Block (CCB). Finally, we validate our approach with extensive experiments on English-Hindi translation tasks in both directions as well as English to Chinese and English to German.</abstract>
<identifier type="citekey">prasad-etal-2025-faster</identifier>
<identifier type="doi">10.18653/v1/2025.findings-naacl.466</identifier>
<location>
<url>https://aclanthology.org/2025.findings-naacl.466/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>8322</start>
<end>8335</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction
%A Prasad, Kritarth
%A Zaki, Mohammadi
%A Singh, Pratik Rakesh
%A Wasnik, Pankaj
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Findings of the Association for Computational Linguistics: NAACL 2025
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-195-7
%F prasad-etal-2025-faster
%X Ensembling neural machine translation (NMT) models to produce higher-quality translations than the L individual models has been extensively studied. Recent methods typically employ a candidate selection block (CSB) and an encoder-decoder fusion block (FB), requiring inference across all candidate models, leading to significant computational overhead, generally Ømega(L). This paper introduces SmartGen, a reinforcement learning (RL)-based strategy that improves the CSB by selecting a small, fixed number of candidates and identifying optimal groups to pass to the fusion block for each input sentence. Furthermore, previously, the CSB and FB were trained independently, leading to suboptimal NMT performance. Our DQN-based SmartGen addresses this by using feedback from the FB block as a reward during training. We also resolve a key issue in earlier methods, where candidates were passed to the FB without modification, by introducing a Competitive Correction Block (CCB). Finally, we validate our approach with extensive experiments on English-Hindi translation tasks in both directions as well as English to Chinese and English to German.
%R 10.18653/v1/2025.findings-naacl.466
%U https://aclanthology.org/2025.findings-naacl.466/
%U https://doi.org/10.18653/v1/2025.findings-naacl.466
%P 8322-8335
Markdown (Informal)
[Faster Machine Translation Ensembling with Reinforcement Learning and Competitive Correction](https://aclanthology.org/2025.findings-naacl.466/) (Prasad et al., Findings 2025)
ACL