Binary Independent Component Analysis: A Non-stationarity-based Approach

We consider independent component analysis of binary data. While fundamental in practice, this case has been much less developed than ICA for continuous data. We start by assuming a linear mixing model in a continuous-valued latent space, followed by a binary observation model. Importantly, we assum...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2022-08
Hauptverfasser: Hyttinen, Antti, Barin-Pacela, Vitória, Hyvärinen, Aapo
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Hyttinen, Antti
Barin-Pacela, Vitória
Hyvärinen, Aapo
description We consider independent component analysis of binary data. While fundamental in practice, this case has been much less developed than ICA for continuous data. We start by assuming a linear mixing model in a continuous-valued latent space, followed by a binary observation model. Importantly, we assume that the sources are non-stationary; this is necessary since any non-Gaussianity would essentially be destroyed by the binarization. Interestingly, the model allows for closed-form likelihood by employing the cumulative distribution function of the multivariate Gaussian distribution. In stark contrast to the continuous-valued case, we prove non-identifiability of the model with few observed variables; our empirical results imply identifiability when the number of observed variables is higher. We present a practical method for binary ICA that uses only pairwise marginals, which are faster to compute than the full multivariate likelihood. Experiments give insight into the requirements for the number of observed variables, segments, and latent sources that allow the model to be estimated.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2605008009</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2605008009</sourcerecordid><originalsourceid>FETCH-proquest_journals_26050080093</originalsourceid><addsrcrecordid>eNqNjLEKwjAYhIMgWLTvEHAO_CZNrW61KHVxci_RRkypSeyfDn17I_gALncH393NSMKF2LAi43xBUsQOAHi-5VKKhNQHY9Uw0bNttddRbKCVe3lnv6m0qp_Q4J6W9OIsw6CCcXFgwsRuCnVLS-8Hp-7PFZk_VI86_fmSrE_Ha1WziN-jxtB0bhziHzY8BwlQAOzEf60PPR076w</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2605008009</pqid></control><display><type>article</type><title>Binary Independent Component Analysis: A Non-stationarity-based Approach</title><source>Free E- Journals</source><creator>Hyttinen, Antti ; Barin-Pacela, Vitória ; Hyvärinen, Aapo</creator><creatorcontrib>Hyttinen, Antti ; Barin-Pacela, Vitória ; Hyvärinen, Aapo</creatorcontrib><description>We consider independent component analysis of binary data. While fundamental in practice, this case has been much less developed than ICA for continuous data. We start by assuming a linear mixing model in a continuous-valued latent space, followed by a binary observation model. Importantly, we assume that the sources are non-stationary; this is necessary since any non-Gaussianity would essentially be destroyed by the binarization. Interestingly, the model allows for closed-form likelihood by employing the cumulative distribution function of the multivariate Gaussian distribution. In stark contrast to the continuous-valued case, we prove non-identifiability of the model with few observed variables; our empirical results imply identifiability when the number of observed variables is higher. We present a practical method for binary ICA that uses only pairwise marginals, which are faster to compute than the full multivariate likelihood. Experiments give insight into the requirements for the number of observed variables, segments, and latent sources that allow the model to be estimated.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Binary data ; Distribution functions ; Empirical analysis ; Independent component analysis ; Multivariate analysis ; Normal distribution</subject><ispartof>arXiv.org, 2022-08</ispartof><rights>2022. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>776,780</link.rule.ids></links><search><creatorcontrib>Hyttinen, Antti</creatorcontrib><creatorcontrib>Barin-Pacela, Vitória</creatorcontrib><creatorcontrib>Hyvärinen, Aapo</creatorcontrib><title>Binary Independent Component Analysis: A Non-stationarity-based Approach</title><title>arXiv.org</title><description>We consider independent component analysis of binary data. While fundamental in practice, this case has been much less developed than ICA for continuous data. We start by assuming a linear mixing model in a continuous-valued latent space, followed by a binary observation model. Importantly, we assume that the sources are non-stationary; this is necessary since any non-Gaussianity would essentially be destroyed by the binarization. Interestingly, the model allows for closed-form likelihood by employing the cumulative distribution function of the multivariate Gaussian distribution. In stark contrast to the continuous-valued case, we prove non-identifiability of the model with few observed variables; our empirical results imply identifiability when the number of observed variables is higher. We present a practical method for binary ICA that uses only pairwise marginals, which are faster to compute than the full multivariate likelihood. Experiments give insight into the requirements for the number of observed variables, segments, and latent sources that allow the model to be estimated.</description><subject>Binary data</subject><subject>Distribution functions</subject><subject>Empirical analysis</subject><subject>Independent component analysis</subject><subject>Multivariate analysis</subject><subject>Normal distribution</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNjLEKwjAYhIMgWLTvEHAO_CZNrW61KHVxci_RRkypSeyfDn17I_gALncH393NSMKF2LAi43xBUsQOAHi-5VKKhNQHY9Uw0bNttddRbKCVe3lnv6m0qp_Q4J6W9OIsw6CCcXFgwsRuCnVLS-8Hp-7PFZk_VI86_fmSrE_Ha1WziN-jxtB0bhziHzY8BwlQAOzEf60PPR076w</recordid><startdate>20220802</startdate><enddate>20220802</enddate><creator>Hyttinen, Antti</creator><creator>Barin-Pacela, Vitória</creator><creator>Hyvärinen, Aapo</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20220802</creationdate><title>Binary Independent Component Analysis: A Non-stationarity-based Approach</title><author>Hyttinen, Antti ; Barin-Pacela, Vitória ; Hyvärinen, Aapo</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26050080093</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Binary data</topic><topic>Distribution functions</topic><topic>Empirical analysis</topic><topic>Independent component analysis</topic><topic>Multivariate analysis</topic><topic>Normal distribution</topic><toplevel>online_resources</toplevel><creatorcontrib>Hyttinen, Antti</creatorcontrib><creatorcontrib>Barin-Pacela, Vitória</creatorcontrib><creatorcontrib>Hyvärinen, Aapo</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Database (Proquest)</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Hyttinen, Antti</au><au>Barin-Pacela, Vitória</au><au>Hyvärinen, Aapo</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Binary Independent Component Analysis: A Non-stationarity-based Approach</atitle><jtitle>arXiv.org</jtitle><date>2022-08-02</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>We consider independent component analysis of binary data. While fundamental in practice, this case has been much less developed than ICA for continuous data. We start by assuming a linear mixing model in a continuous-valued latent space, followed by a binary observation model. Importantly, we assume that the sources are non-stationary; this is necessary since any non-Gaussianity would essentially be destroyed by the binarization. Interestingly, the model allows for closed-form likelihood by employing the cumulative distribution function of the multivariate Gaussian distribution. In stark contrast to the continuous-valued case, we prove non-identifiability of the model with few observed variables; our empirical results imply identifiability when the number of observed variables is higher. We present a practical method for binary ICA that uses only pairwise marginals, which are faster to compute than the full multivariate likelihood. Experiments give insight into the requirements for the number of observed variables, segments, and latent sources that allow the model to be estimated.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2022-08
issn 2331-8422
language eng
recordid cdi_proquest_journals_2605008009
source Free E- Journals
subjects Binary data
Distribution functions
Empirical analysis
Independent component analysis
Multivariate analysis
Normal distribution
title Binary Independent Component Analysis: A Non-stationarity-based Approach
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-01T10%3A51%3A55IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Binary%20Independent%20Component%20Analysis:%20A%20Non-stationarity-based%20Approach&rft.jtitle=arXiv.org&rft.au=Hyttinen,%20Antti&rft.date=2022-08-02&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2605008009%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2605008009&rft_id=info:pmid/&rfr_iscdi=true