Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition

Tanmoy  Mukherjee; Timothy Hospedales

doi:10.18653/v1/D16-1089

Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

An exciting outcome of research at the intersection of language and vision is that of zeroshot learning (ZSL). ZSL promises to scale visual recognition by borrowing distributed semantic models learned from linguistic corpora and turning them into visual recognition models. However the popular word-vector DSM embeddings are relatively impoverished in their expressivity as they model each word as a single vector point. In this paper we explore word-distribution embeddings for ZSL. We present a visual-linguistic mapping for ZSL in the case where words and visual categories are both represented by distributions. Experiments show improved results on ZSL benchmarks due to this better exploiting of intra-concept variability in each modality

Original language	English
Title of host publication	Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing
Publisher	Association for Computational Linguistics (ACL)
Pages	912-918
Number of pages	7
ISBN (Print)	978-1-945626-25-8
DOIs	https://doi.org/10.18653/v1/D16-1089
Publication status	Published - 5 Nov 2016
Event	2016 Conference on Empirical Methods in Natural Language Processing - Austin, United States Duration: 1 Nov 2016 → 5 Nov 2016 https://www.aclweb.org/mirror/emnlp2016/

Conference

Conference	2016 Conference on Empirical Methods in Natural Language Processing
Abbreviated title	EMNLP 2016
Country/Territory	United States
City	Austin
Period	1/11/16 → 5/11/16
Internet address	https://www.aclweb.org/mirror/emnlp2016/

Access to Document

10.18653/v1/D16-1089Licence: Creative Commons: Attribution (CC-BY)

D16-1089Final published version, 2.99 MB

https://www.aclweb.org/anthology/D/D16/D16-1089

Cite this

@inproceedings{bfef9f6d617c4e7a88cf4f8ab8404e97,

title = "Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition",

abstract = "An exciting outcome of research at the intersection of language and vision is that of zeroshot learning (ZSL). ZSL promises to scale visual recognition by borrowing distributed semantic models learned from linguistic corpora and turning them into visual recognition models. However the popular word-vector DSM embeddings are relatively impoverished in their expressivity as they model each word as a single vector point. In this paper we explore word-distribution embeddings for ZSL. We present a visual-linguistic mapping for ZSL in the case where words and visual categories are both represented by distributions. Experiments show improved results on ZSL benchmarks due to this better exploiting of intra-concept variability in each modality",

author = "Tanmoy Mukherjee and Timothy Hospedales",

year = "2016",

month = nov,

day = "5",

doi = "10.18653/v1/D16-1089",

language = "English",

isbn = "978-1-945626-25-8",

pages = "912--918",

booktitle = "Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing",

publisher = "Association for Computational Linguistics (ACL)",

note = "2016 Conference on Empirical Methods in Natural Language Processing, EMNLP 2016 ; Conference date: 01-11-2016 Through 05-11-2016",

url = "https://www.aclweb.org/mirror/emnlp2016/",

}

Mukherjee, T & Hospedales, T 2016, Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition. in Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics (ACL), pp. 912-918, 2016 Conference on Empirical Methods in Natural Language Processing, Austin, Texas, United States, 1/11/16. https://doi.org/10.18653/v1/D16-1089

TY - GEN

T1 - Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition

AU - Mukherjee, Tanmoy

AU - Hospedales, Timothy

PY - 2016/11/5

Y1 - 2016/11/5

N2 - An exciting outcome of research at the intersection of language and vision is that of zeroshot learning (ZSL). ZSL promises to scale visual recognition by borrowing distributed semantic models learned from linguistic corpora and turning them into visual recognition models. However the popular word-vector DSM embeddings are relatively impoverished in their expressivity as they model each word as a single vector point. In this paper we explore word-distribution embeddings for ZSL. We present a visual-linguistic mapping for ZSL in the case where words and visual categories are both represented by distributions. Experiments show improved results on ZSL benchmarks due to this better exploiting of intra-concept variability in each modality

AB - An exciting outcome of research at the intersection of language and vision is that of zeroshot learning (ZSL). ZSL promises to scale visual recognition by borrowing distributed semantic models learned from linguistic corpora and turning them into visual recognition models. However the popular word-vector DSM embeddings are relatively impoverished in their expressivity as they model each word as a single vector point. In this paper we explore word-distribution embeddings for ZSL. We present a visual-linguistic mapping for ZSL in the case where words and visual categories are both represented by distributions. Experiments show improved results on ZSL benchmarks due to this better exploiting of intra-concept variability in each modality

UR - http://dblp2.uni-trier.de/db/conf/emnlp/emnlp2016.html

U2 - 10.18653/v1/D16-1089

DO - 10.18653/v1/D16-1089

M3 - Conference contribution

SN - 978-1-945626-25-8

SP - 912

EP - 918

BT - Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing

PB - Association for Computational Linguistics (ACL)

T2 - 2016 Conference on Empirical Methods in Natural Language Processing

Y2 - 1 November 2016 through 5 November 2016

ER -

Gaussian Visual-Linguistic Embedding for Zero-Shot Recognition

Abstract / Description of output

Conference

Access to Document

Other files and links

Fingerprint

Cite this