Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval

Victor Petren Bach Hansen; Anders Sogaard

doi:10.1109/SITIS.2018.00072

Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval

Victor Petren Bach Hansen, Anders Sogaard

Datalogisk Institut

Abstract

This paper presents a simple modification to previous work on learning cross-lingual, grounded word representations from image-word pairs that, unlike previous work, is robust across different parts of speech, e.g., able to find the translation of the adjective 'social' relying only on image features associated with its translation candidates. Our method does not rely on black-box image search engines or any direct cross-lingual supervision. We evaluate our approach on English-German and English-Japanese word alignment, as well as on existing English-German bilingual dictionary induction datasets.

Originalsprog	Engelsk
Titel	Proceedings - 14th International Conference on Signal-Image Technology and Internet Based Systems, SITIS
Antal sider	8
Forlag	IEEE
Publikationsdato	2 jul. 2018
Sider	427-434
ISBN (Elektronisk)	978-1-5386-9385-8
DOI	https://doi.org/10.1109/SITIS.2018.00072
Status	Udgivet - 2 jul. 2018
Begivenhed	14th International Conference on Signal Image Technology and Internet Based Systems, SITIS 2018 - Las Palmas de Gran Canaria, Spanien Varighed: 26 nov. 2018 → 29 nov. 2018

Konference

Konference	14th International Conference on Signal Image Technology and Internet Based Systems, SITIS 2018
Land/Område	Spanien
By	Las Palmas de Gran Canaria
Periode	26/11/2018 → 29/11/2018

Adgang til dokumentet

10.1109/SITIS.2018.00072

Andre filer og links

Link to publication in Scopus

Citationsformater

Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval. / Hansen, Victor Petren Bach ; Sogaard, Anders.
Proceedings - 14th International Conference on Signal-Image Technology and Internet Based Systems, SITIS. IEEE, 2018. s. 427-434.

Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › peer review

Hansen, VPB & Sogaard, A 2018, Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval. i Proceedings - 14th International Conference on Signal-Image Technology and Internet Based Systems, SITIS. IEEE, s. 427-434, 14th International Conference on Signal Image Technology and Internet Based Systems, SITIS 2018, Las Palmas de Gran Canaria, Spanien, 26/11/2018. https://doi.org/10.1109/SITIS.2018.00072

@inproceedings{00a0e191d4864d6dbf69dc26676c02fb,

title = "Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval",

abstract = "This paper presents a simple modification to previous work on learning cross-lingual, grounded word representations from image-word pairs that, unlike previous work, is robust across different parts of speech, e.g., able to find the translation of the adjective 'social' relying only on image features associated with its translation candidates. Our method does not rely on black-box image search engines or any direct cross-lingual supervision. We evaluate our approach on English-German and English-Japanese word alignment, as well as on existing English-German bilingual dictionary induction datasets.",

keywords = "Computer vision, Cross-lingual learning, Distributional semantics, Multi-modal retrieval, Natural language processing",

author = "Hansen, {Victor Petren Bach} and Anders Sogaard",

year = "2018",

month = jul,

day = "2",

doi = "10.1109/SITIS.2018.00072",

language = "English",

pages = "427--434",

booktitle = "Proceedings - 14th International Conference on Signal-Image Technology and Internet Based Systems, SITIS",

publisher = "IEEE",

note = "14th International Conference on Signal Image Technology and Internet Based Systems, SITIS 2018 ; Conference date: 26-11-2018 Through 29-11-2018",

}

TY - GEN

T1 - Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval

AU - Hansen, Victor Petren Bach

AU - Sogaard, Anders

PY - 2018/7/2

Y1 - 2018/7/2

N2 - This paper presents a simple modification to previous work on learning cross-lingual, grounded word representations from image-word pairs that, unlike previous work, is robust across different parts of speech, e.g., able to find the translation of the adjective 'social' relying only on image features associated with its translation candidates. Our method does not rely on black-box image search engines or any direct cross-lingual supervision. We evaluate our approach on English-German and English-Japanese word alignment, as well as on existing English-German bilingual dictionary induction datasets.

AB - This paper presents a simple modification to previous work on learning cross-lingual, grounded word representations from image-word pairs that, unlike previous work, is robust across different parts of speech, e.g., able to find the translation of the adjective 'social' relying only on image features associated with its translation candidates. Our method does not rely on black-box image search engines or any direct cross-lingual supervision. We evaluate our approach on English-German and English-Japanese word alignment, as well as on existing English-German bilingual dictionary induction datasets.

KW - Computer vision

KW - Cross-lingual learning

KW - Distributional semantics

KW - Multi-modal retrieval

KW - Natural language processing

UR - http://www.scopus.com/inward/record.url?scp=85065904244&partnerID=8YFLogxK

U2 - 10.1109/SITIS.2018.00072

DO - 10.1109/SITIS.2018.00072

M3 - Article in proceedings

AN - SCOPUS:85065904244

SP - 427

EP - 434

BT - Proceedings - 14th International Conference on Signal-Image Technology and Internet Based Systems, SITIS

PB - IEEE

T2 - 14th International Conference on Signal Image Technology and Internet Based Systems, SITIS 2018

Y2 - 26 November 2018 through 29 November 2018

ER -

Learning Language-Independent Representations of Verbs and Adjectives from Multimodal Retrieval

Abstract

Konference

Adgang til dokumentet

Andre filer og links

Fingeraftryk

Citationsformater