<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nguyen-etal-2021-skim-attention">
    <titleInfo>
        <title>Skim-Attention: Learning to Focus via Document Layout</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Laura</namePart>
        <namePart type="family">Nguyen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Thomas</namePart>
        <namePart type="family">Scialom</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jacopo</namePart>
        <namePart type="family">Staiano</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Benjamin</namePart>
        <namePart type="family">Piwowarski</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2021-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: EMNLP 2021</title>
        </titleInfo>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Punta Cana, Dominican Republic</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Transformer-based pre-training techniques of text and layout have proven effective in a number of document understanding tasks. Despite this success, multimodal pre-training models suffer from very high computational and memory costs. Motivated by human reading strategies, this paper presents Skim-Attention, a new attention mechanism that takes advantage of the structure of the document and its layout. Skim-Attention only attends to the 2-dimensional position of the words in a document. Our experiments show that Skim-Attention obtains a lower perplexity than prior works, while being more computationally efficient. Skim-Attention can be further combined with long-range Transformers to efficiently process long documents. We also show how Skim-Attention can be used off-the-shelf as a mask for any Pre-trained Language Model, allowing to improve their performance while restricting attention. Finally, we show the emergence of a document structure representation in Skim-Attention.</abstract>
    <identifier type="citekey">nguyen-etal-2021-skim-attention</identifier>
    <identifier type="doi">10.18653/v1/2021.findings-emnlp.207</identifier>
    <location>
        <url>https://aclanthology.org/2021.findings-emnlp.207</url>
    </location>
    <part>
        <date>2021-11</date>
        <extent unit="page">
            <start>2413</start>
            <end>2427</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Skim-Attention: Learning to Focus via Document Layout
%A Nguyen, Laura
%A Scialom, Thomas
%A Staiano, Jacopo
%A Piwowarski, Benjamin
%S Findings of the Association for Computational Linguistics: EMNLP 2021
%D 2021
%8 November
%I Association for Computational Linguistics
%C Punta Cana, Dominican Republic
%F nguyen-etal-2021-skim-attention
%X Transformer-based pre-training techniques of text and layout have proven effective in a number of document understanding tasks. Despite this success, multimodal pre-training models suffer from very high computational and memory costs. Motivated by human reading strategies, this paper presents Skim-Attention, a new attention mechanism that takes advantage of the structure of the document and its layout. Skim-Attention only attends to the 2-dimensional position of the words in a document. Our experiments show that Skim-Attention obtains a lower perplexity than prior works, while being more computationally efficient. Skim-Attention can be further combined with long-range Transformers to efficiently process long documents. We also show how Skim-Attention can be used off-the-shelf as a mask for any Pre-trained Language Model, allowing to improve their performance while restricting attention. Finally, we show the emergence of a document structure representation in Skim-Attention.
%R 10.18653/v1/2021.findings-emnlp.207
%U https://aclanthology.org/2021.findings-emnlp.207
%U https://doi.org/10.18653/v1/2021.findings-emnlp.207
%P 2413-2427

Markdown (Informal)

[Skim-Attention: Learning to Focus via Document Layout](https://aclanthology.org/2021.findings-emnlp.207) (Nguyen et al., Findings 2021)

• Skim-Attention: Learning to Focus via Document Layout (Nguyen et al., Findings 2021)

ACL

• Laura Nguyen, Thomas Scialom, Jacopo Staiano, and Benjamin Piwowarski. 2021. Skim-Attention: Learning to Focus via Document Layout. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 2413–2427, Punta Cana, Dominican Republic. Association for Computational Linguistics.