Sökning: id:"swepub:oai:DiVA.org:uu-457404" >
Bidirectional Domai...
-
Dakota, DanielUppsala universitet,Institutionen för lingvistik och filologi
(författare)
Bidirectional Domain Adaptation Using Weighted Multi-Task Learning
- Artikel/kapitelEngelska2021
Förlag, utgivningsår, omfång ...
-
Stroudsburg, PA, USA :Association for Computational Linguistics,2021
-
printrdacarrier
Nummerbeteckningar
-
LIBRIS-ID:oai:DiVA.org:uu-457404
-
https://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-457404URI
-
https://doi.org/10.18653/v1/2021.iwpt-1.10DOI
Kompletterande språkuppgifter
-
Språk:engelska
-
Sammanfattning på:engelska
Ingår i deldatabas
Klassifikation
-
Ämneskategori:ref swepub-contenttype
-
Ämneskategori:kon swepub-publicationtype
Anmärkningar
-
Domain adaption in syntactic parsing is still a significant challenge. We address the issue of data imbalance between the in-domain and out-of-domain treebank typically used for the problem. We define domain adaptation as a Multi-task learning (MTL) problem, which allows us to train two parsers, one for each domain. Our results show that the MTL approach is beneficial for the smaller treebank. For the larger treebank, we need to use loss weighting in order to avoid a decrease in performance below the single task. In order to determine to what degree the data imbalance between two domains and the domain differences affect results, we also carry out an experiment with two imbalanced in-domain treebanks and show that loss weighting also improves performance in an in-domain setting. Given loss weighting in MTL, we can improve results for both parsers.
Ämnesord och genrebeteckningar
Biuppslag (personer, institutioner, konferenser, titlar ...)
-
Sayyed, Zeeshan AliIndiana Univ, Bloomington, IN 47405 USA.
(författare)
-
Kuebler, SandraIndiana Univ, Bloomington, IN 47405 USA.
(författare)
-
Uppsala universitetInstitutionen för lingvistik och filologi
(creator_code:org_t)
Sammanhörande titlar
-
Ingår i:IWPT 2021Stroudsburg, PA, USA : Association for Computational Linguistics, s. 93-1059781954085800
Internetlänk
Hitta via bibliotek
Till lärosätets databas