Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity

Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity

This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, re...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Eronen, Juuso, Ptaszynski, Michal, Nowakowski, Karol, Chia, Zheng Lin, Masui, Fumito
Format: Artikel
Sprache:eng
Schlagworte:
Computer Science - Computation and Language
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Eronen, Juuso
Ptaszynski, Michal
Nowakowski, Karol
Chia, Zheng Lin
Masui, Fumito
description This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, related languages can also be particularly effective when there is limited data available for a specific language pair, as the model can leverage the similarities between the languages to improve performance. To demonstrate, we fine-tune mBART model for a Polish-English translation task using the OPUS-100 dataset. We evaluate the performance of the model under various transfer learning configurations, including different transfer source languages and different shot levels for Polish, and report the results. Our experiments show that a combination of related languages and larger amounts of data outperforms the model trained on related languages or larger amounts of data alone. Additionally, we show the importance of related languages in zero-shot and few-shot configurations.
doi_str_mv 10.48550/arxiv.2306.00660
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2306_00660</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2306_00660</sourcerecordid><originalsourceid>FETCH-LOGICAL-a670-c068a99e5b57d771caaa249c0a0889eda5402962e1313144f86d3fad2a7f98003</originalsourceid><addsrcrecordid>eNotkMtOwzAURL1hgQofwIr7Awk3Lydmh0qASuEhEbGNLomdWHLsykkKXfDvlBbNYkazONIMY1cRhmmRZXhD_lvvwjhBHiJyjufsZzNuvdtp28ObM3oaYHZQ2v4YX-TiycAztYO2EmpPdjI0a2fhS8_DqVDSQyXJ2wPjFkqlZDtP4BTc00zw4cwySiDbQUW2X6iX8K5HbcjreX_BzhSZSV7--4rVD2W9fgqq18fN-q4KiOcYtMgLEkJmn1ne5XnUElGcihYJi0LIjrIUY8FjGSUHpakqeJco6mLKlSgQkxW7PmGP-5ut1yP5ffP3Q3P8IfkF0u5Zgg</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity</title><source>arXiv.org</source><creator>Eronen, Juuso ; Ptaszynski, Michal ; Nowakowski, Karol ; Chia, Zheng Lin ; Masui, Fumito</creator><creatorcontrib>Eronen, Juuso ; Ptaszynski, Michal ; Nowakowski, Karol ; Chia, Zheng Lin ; Masui, Fumito</creatorcontrib><description>This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, related languages can also be particularly effective when there is limited data available for a specific language pair, as the model can leverage the similarities between the languages to improve performance. To demonstrate, we fine-tune mBART model for a Polish-English translation task using the OPUS-100 dataset. We evaluate the performance of the model under various transfer learning configurations, including different transfer source languages and different shot levels for Polish, and report the results. Our experiments show that a combination of related languages and larger amounts of data outperforms the model trained on related languages or larger amounts of data alone. Additionally, we show the importance of related languages in zero-shot and few-shot configurations.</description><identifier>DOI: 10.48550/arxiv.2306.00660</identifier><language>eng</language><subject>Computer Science - Computation and Language</subject><creationdate>2023-06</creationdate><rights>http://creativecommons.org/licenses/by-nc-nd/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2306.00660$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2306.00660$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Eronen, Juuso</creatorcontrib><creatorcontrib>Ptaszynski, Michal</creatorcontrib><creatorcontrib>Nowakowski, Karol</creatorcontrib><creatorcontrib>Chia, Zheng Lin</creatorcontrib><creatorcontrib>Masui, Fumito</creatorcontrib><title>Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity</title><description>This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, related languages can also be particularly effective when there is limited data available for a specific language pair, as the model can leverage the similarities between the languages to improve performance. To demonstrate, we fine-tune mBART model for a Polish-English translation task using the OPUS-100 dataset. We evaluate the performance of the model under various transfer learning configurations, including different transfer source languages and different shot levels for Polish, and report the results. Our experiments show that a combination of related languages and larger amounts of data outperforms the model trained on related languages or larger amounts of data alone. Additionally, we show the importance of related languages in zero-shot and few-shot configurations.</description><subject>Computer Science - Computation and Language</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotkMtOwzAURL1hgQofwIr7Awk3Lydmh0qASuEhEbGNLomdWHLsykkKXfDvlBbNYkazONIMY1cRhmmRZXhD_lvvwjhBHiJyjufsZzNuvdtp28ObM3oaYHZQ2v4YX-TiycAztYO2EmpPdjI0a2fhS8_DqVDSQyXJ2wPjFkqlZDtP4BTc00zw4cwySiDbQUW2X6iX8K5HbcjreX_BzhSZSV7--4rVD2W9fgqq18fN-q4KiOcYtMgLEkJmn1ne5XnUElGcihYJi0LIjrIUY8FjGSUHpakqeJco6mLKlSgQkxW7PmGP-5ut1yP5ffP3Q3P8IfkF0u5Zgg</recordid><startdate>20230601</startdate><enddate>20230601</enddate><creator>Eronen, Juuso</creator><creator>Ptaszynski, Michal</creator><creator>Nowakowski, Karol</creator><creator>Chia, Zheng Lin</creator><creator>Masui, Fumito</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230601</creationdate><title>Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity</title><author>Eronen, Juuso ; Ptaszynski, Michal ; Nowakowski, Karol ; Chia, Zheng Lin ; Masui, Fumito</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a670-c068a99e5b57d771caaa249c0a0889eda5402962e1313144f86d3fad2a7f98003</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Computation and Language</topic><toplevel>online_resources</toplevel><creatorcontrib>Eronen, Juuso</creatorcontrib><creatorcontrib>Ptaszynski, Michal</creatorcontrib><creatorcontrib>Nowakowski, Karol</creatorcontrib><creatorcontrib>Chia, Zheng Lin</creatorcontrib><creatorcontrib>Masui, Fumito</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Eronen, Juuso</au><au>Ptaszynski, Michal</au><au>Nowakowski, Karol</au><au>Chia, Zheng Lin</au><au>Masui, Fumito</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity</atitle><date>2023-06-01</date><risdate>2023</risdate><abstract>This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, related languages can also be particularly effective when there is limited data available for a specific language pair, as the model can leverage the similarities between the languages to improve performance. To demonstrate, we fine-tune mBART model for a Polish-English translation task using the OPUS-100 dataset. We evaluate the performance of the model under various transfer learning configurations, including different transfer source languages and different shot levels for Polish, and report the results. Our experiments show that a combination of related languages and larger amounts of data outperforms the model trained on related languages or larger amounts of data alone. Additionally, we show the importance of related languages in zero-shot and few-shot configurations.</abstract><doi>10.48550/arxiv.2306.00660</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2306.00660
ispartof
issn
language eng
recordid cdi_arxiv_primary_2306_00660
source arXiv.org
subjects Computer Science - Computation and Language
title Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-02T22%3A36%3A21IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Improving%20Polish%20to%20English%20Neural%20Machine%20Translation%20with%20Transfer%20Learning:%20Effects%20of%20Data%20Volume%20and%20Language%20Similarity&rft.au=Eronen,%20Juuso&rft.date=2023-06-01&rft_id=info:doi/10.48550/arxiv.2306.00660&rft_dat=%3Carxiv_GOX%3E2306_00660%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true