@inproceedings{ilharco-etal-2019-large,
title = "Large-Scale Representation Learning from Visually Grounded Untranscribed Speech",
author = "Ilharco, Gabriel and
Zhang, Yuan and
Baldridge, Jason",
editor = "Bansal, Mohit and
Villavicencio, Aline",
booktitle = "Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)",
month = nov,
year = "2019",
address = "Hong Kong, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/K19-1006",
doi = "10.18653/v1/K19-1006",
pages = "55--65",
abstract = "Systems that can associate images with their spoken audio captions are an important step towards visually grounded language learning. We describe a scalable method to automatically generate diverse audio for image captioning datasets. This supports pretraining deep networks for encoding both audio and images, which we do via a dual encoder that learns to align latent representations from both modalities. We show that a masked margin softmax loss for such models is superior to the standard triplet loss. We fine-tune these models on the Flickr8k Audio Captions Corpus and obtain state-of-the-art results{---}improving recall in the top 10 from 29.6{\%} to 49.5{\%}. We also obtain human ratings on retrieval outputs to better assess the impact of incidentally matching image-caption pairs that were not associated in the data, finding that automatic evaluation substantially underestimates the quality of the retrieved results.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ilharco-etal-2019-large">
<titleInfo>
<title>Large-Scale Representation Learning from Visually Grounded Untranscribed Speech</title>
</titleInfo>
<name type="personal">
<namePart type="given">Gabriel</namePart>
<namePart type="family">Ilharco</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jason</namePart>
<namePart type="family">Baldridge</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2019-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mohit</namePart>
<namePart type="family">Bansal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aline</namePart>
<namePart type="family">Villavicencio</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hong Kong, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Systems that can associate images with their spoken audio captions are an important step towards visually grounded language learning. We describe a scalable method to automatically generate diverse audio for image captioning datasets. This supports pretraining deep networks for encoding both audio and images, which we do via a dual encoder that learns to align latent representations from both modalities. We show that a masked margin softmax loss for such models is superior to the standard triplet loss. We fine-tune these models on the Flickr8k Audio Captions Corpus and obtain state-of-the-art results—improving recall in the top 10 from 29.6% to 49.5%. We also obtain human ratings on retrieval outputs to better assess the impact of incidentally matching image-caption pairs that were not associated in the data, finding that automatic evaluation substantially underestimates the quality of the retrieved results.</abstract>
<identifier type="citekey">ilharco-etal-2019-large</identifier>
<identifier type="doi">10.18653/v1/K19-1006</identifier>
<location>
<url>https://aclanthology.org/K19-1006</url>
</location>
<part>
<date>2019-11</date>
<extent unit="page">
<start>55</start>
<end>65</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Large-Scale Representation Learning from Visually Grounded Untranscribed Speech
%A Ilharco, Gabriel
%A Zhang, Yuan
%A Baldridge, Jason
%Y Bansal, Mohit
%Y Villavicencio, Aline
%S Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)
%D 2019
%8 November
%I Association for Computational Linguistics
%C Hong Kong, China
%F ilharco-etal-2019-large
%X Systems that can associate images with their spoken audio captions are an important step towards visually grounded language learning. We describe a scalable method to automatically generate diverse audio for image captioning datasets. This supports pretraining deep networks for encoding both audio and images, which we do via a dual encoder that learns to align latent representations from both modalities. We show that a masked margin softmax loss for such models is superior to the standard triplet loss. We fine-tune these models on the Flickr8k Audio Captions Corpus and obtain state-of-the-art results—improving recall in the top 10 from 29.6% to 49.5%. We also obtain human ratings on retrieval outputs to better assess the impact of incidentally matching image-caption pairs that were not associated in the data, finding that automatic evaluation substantially underestimates the quality of the retrieved results.
%R 10.18653/v1/K19-1006
%U https://aclanthology.org/K19-1006
%U https://doi.org/10.18653/v1/K19-1006
%P 55-65
Markdown (Informal)
[Large-Scale Representation Learning from Visually Grounded Untranscribed Speech](https://aclanthology.org/K19-1006) (Ilharco et al., CoNLL 2019)
ACL