<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hiebert-etal-2018-interpreting">
    <titleInfo>
        <title>Interpreting Word-Level Hidden State Behaviour of Character-Level LSTM Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Avery</namePart>
        <namePart type="family">Hiebert</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Cole</namePart>
        <namePart type="family">Peterson</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Alona</namePart>
        <namePart type="family">Fyshe</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Nishant</namePart>
        <namePart type="family">Mehta</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2018-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Tal</namePart>
            <namePart type="family">Linzen</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Grzegorz</namePart>
            <namePart type="family">Chrupała</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Afra</namePart>
            <namePart type="family">Alishahi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Brussels, Belgium</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>While Long Short-Term Memory networks (LSTMs) and other forms of recurrent neural network have been successfully applied to language modeling on a character level, the hidden state dynamics of these models can be difficult to interpret. We investigate the hidden states of such a model by using the HDBSCAN clustering algorithm to identify points in the text at which the hidden state is similar. Focusing on whitespace characters prior to the beginning of a word reveals interpretable clusters that offer insight into how the LSTM may combine contextual and character-level information to identify parts of speech. We also introduce a method for deriving word vectors from the hidden state representation in order to investigate the word-level knowledge of the model. These word vectors encode meaningful semantic information even for words that appear only once in the training text.</abstract>
    <identifier type="citekey">hiebert-etal-2018-interpreting</identifier>
    <identifier type="doi">10.18653/v1/W18-5428</identifier>
    <location>
        <url>https://aclanthology.org/W18-5428</url>
    </location>
    <part>
        <date>2018-11</date>
        <extent unit="page">
            <start>258</start>
            <end>266</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Interpreting Word-Level Hidden State Behaviour of Character-Level LSTM Language Models
%A Hiebert, Avery
%A Peterson, Cole
%A Fyshe, Alona
%A Mehta, Nishant
%Y Linzen, Tal
%Y Chrupała, Grzegorz
%Y Alishahi, Afra
%S Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP
%D 2018
%8 November
%I Association for Computational Linguistics
%C Brussels, Belgium
%F hiebert-etal-2018-interpreting
%X While Long Short-Term Memory networks (LSTMs) and other forms of recurrent neural network have been successfully applied to language modeling on a character level, the hidden state dynamics of these models can be difficult to interpret. We investigate the hidden states of such a model by using the HDBSCAN clustering algorithm to identify points in the text at which the hidden state is similar. Focusing on whitespace characters prior to the beginning of a word reveals interpretable clusters that offer insight into how the LSTM may combine contextual and character-level information to identify parts of speech. We also introduce a method for deriving word vectors from the hidden state representation in order to investigate the word-level knowledge of the model. These word vectors encode meaningful semantic information even for words that appear only once in the training text.
%R 10.18653/v1/W18-5428
%U https://aclanthology.org/W18-5428
%U https://doi.org/10.18653/v1/W18-5428
%P 258-266

Markdown (Informal)

[Interpreting Word-Level Hidden State Behaviour of Character-Level LSTM Language Models](https://aclanthology.org/W18-5428) (Hiebert et al., EMNLP 2018)

• Interpreting Word-Level Hidden State Behaviour of Character-Level LSTM Language Models (Hiebert et al., EMNLP 2018)

ACL

• Avery Hiebert, Cole Peterson, Alona Fyshe, and Nishant Mehta. 2018. Interpreting Word-Level Hidden State Behaviour of Character-Level LSTM Language Models. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pages 258–266, Brussels, Belgium. Association for Computational Linguistics.