FAVAE: Sequence Disentanglement using Information Bottleneck Principle

We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yamada, Masanori, Kim, Heecheol, Miyoshi, Kosuke, Yamakawa, Hiroshi
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Yamada, Masanori
Kim, Heecheol
Miyoshi, Kosuke
Yamakawa, Hiroshi
description We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain interpretable and transferable representations from data. We focused on the disentangled representation of sequential data since there is a wide range of potential applications if disentanglement representation is extended to sequential data such as video, speech, and stock market. Sequential data are characterized by dynamic and static factors: dynamic factors are time dependent, and static factors are independent of time. Previous models disentangle static and dynamic factors by explicitly modeling the priors of latent variables to distinguish between these factors. However, these models cannot disentangle representations between dynamic factors, such as disentangling "picking up" and "throwing" in robotic tasks. FAVAE can disentangle multiple dynamic factors. Since it does not require modeling priors, it can disentangle "between" dynamic factors. We conducted experiments to show that FAVAE can extract disentangled dynamic factors.
doi_str_mv 10.48550/arxiv.1902.08341
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1902_08341</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1902_08341</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-4162cf0dd6e9a1fef3128b5b320f2da8f471a2b61a76920f34f5f5bbf01b681d3</originalsourceid><addsrcrecordid>eNotz01OwzAUBGBvWKDCAVjhCyT4-S8Ou1AaqFQJJCq2kZ08VxaJU5wUwe1pC6sZzWKkj5AbYLk0SrE7m77DVw4l4zkzQsIlqevqvVrd0zf8PGBskT6GCeNs467H4VjoYQpxR9fRj2mwcxgjfRjnuceI7Qd9TSG2Yd_jFbnwtp_w-j8XZFuvtsvnbPPytF5Wm8zqAjIJmreedZ3G0oJHL4Abp5zgzPPOGi8LsNxpsIUuj5uQXnnlnGfgtIFOLMjt3-0Z0uxTGGz6aU6g5gwSv-qPRiU</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>FAVAE: Sequence Disentanglement using Information Bottleneck Principle</title><source>arXiv.org</source><creator>Yamada, Masanori ; Kim, Heecheol ; Miyoshi, Kosuke ; Yamakawa, Hiroshi</creator><creatorcontrib>Yamada, Masanori ; Kim, Heecheol ; Miyoshi, Kosuke ; Yamakawa, Hiroshi</creatorcontrib><description>We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain interpretable and transferable representations from data. We focused on the disentangled representation of sequential data since there is a wide range of potential applications if disentanglement representation is extended to sequential data such as video, speech, and stock market. Sequential data are characterized by dynamic and static factors: dynamic factors are time dependent, and static factors are independent of time. Previous models disentangle static and dynamic factors by explicitly modeling the priors of latent variables to distinguish between these factors. However, these models cannot disentangle representations between dynamic factors, such as disentangling "picking up" and "throwing" in robotic tasks. FAVAE can disentangle multiple dynamic factors. Since it does not require modeling priors, it can disentangle "between" dynamic factors. We conducted experiments to show that FAVAE can extract disentangled dynamic factors.</description><identifier>DOI: 10.48550/arxiv.1902.08341</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2019-02</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1902.08341$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1902.08341$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Yamada, Masanori</creatorcontrib><creatorcontrib>Kim, Heecheol</creatorcontrib><creatorcontrib>Miyoshi, Kosuke</creatorcontrib><creatorcontrib>Yamakawa, Hiroshi</creatorcontrib><title>FAVAE: Sequence Disentanglement using Information Bottleneck Principle</title><description>We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain interpretable and transferable representations from data. We focused on the disentangled representation of sequential data since there is a wide range of potential applications if disentanglement representation is extended to sequential data such as video, speech, and stock market. Sequential data are characterized by dynamic and static factors: dynamic factors are time dependent, and static factors are independent of time. Previous models disentangle static and dynamic factors by explicitly modeling the priors of latent variables to distinguish between these factors. However, these models cannot disentangle representations between dynamic factors, such as disentangling "picking up" and "throwing" in robotic tasks. FAVAE can disentangle multiple dynamic factors. Since it does not require modeling priors, it can disentangle "between" dynamic factors. We conducted experiments to show that FAVAE can extract disentangled dynamic factors.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz01OwzAUBGBvWKDCAVjhCyT4-S8Ou1AaqFQJJCq2kZ08VxaJU5wUwe1pC6sZzWKkj5AbYLk0SrE7m77DVw4l4zkzQsIlqevqvVrd0zf8PGBskT6GCeNs467H4VjoYQpxR9fRj2mwcxgjfRjnuceI7Qd9TSG2Yd_jFbnwtp_w-j8XZFuvtsvnbPPytF5Wm8zqAjIJmreedZ3G0oJHL4Abp5zgzPPOGi8LsNxpsIUuj5uQXnnlnGfgtIFOLMjt3-0Z0uxTGGz6aU6g5gwSv-qPRiU</recordid><startdate>20190221</startdate><enddate>20190221</enddate><creator>Yamada, Masanori</creator><creator>Kim, Heecheol</creator><creator>Miyoshi, Kosuke</creator><creator>Yamakawa, Hiroshi</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20190221</creationdate><title>FAVAE: Sequence Disentanglement using Information Bottleneck Principle</title><author>Yamada, Masanori ; Kim, Heecheol ; Miyoshi, Kosuke ; Yamakawa, Hiroshi</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-4162cf0dd6e9a1fef3128b5b320f2da8f471a2b61a76920f34f5f5bbf01b681d3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Yamada, Masanori</creatorcontrib><creatorcontrib>Kim, Heecheol</creatorcontrib><creatorcontrib>Miyoshi, Kosuke</creatorcontrib><creatorcontrib>Yamakawa, Hiroshi</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Yamada, Masanori</au><au>Kim, Heecheol</au><au>Miyoshi, Kosuke</au><au>Yamakawa, Hiroshi</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>FAVAE: Sequence Disentanglement using Information Bottleneck Principle</atitle><date>2019-02-21</date><risdate>2019</risdate><abstract>We propose the factorized action variational autoencoder (FAVAE), a state-of-the-art generative model for learning disentangled and interpretable representations from sequential data via the information bottleneck without supervision. The purpose of disentangled representation learning is to obtain interpretable and transferable representations from data. We focused on the disentangled representation of sequential data since there is a wide range of potential applications if disentanglement representation is extended to sequential data such as video, speech, and stock market. Sequential data are characterized by dynamic and static factors: dynamic factors are time dependent, and static factors are independent of time. Previous models disentangle static and dynamic factors by explicitly modeling the priors of latent variables to distinguish between these factors. However, these models cannot disentangle representations between dynamic factors, such as disentangling "picking up" and "throwing" in robotic tasks. FAVAE can disentangle multiple dynamic factors. Since it does not require modeling priors, it can disentangle "between" dynamic factors. We conducted experiments to show that FAVAE can extract disentangled dynamic factors.</abstract><doi>10.48550/arxiv.1902.08341</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1902.08341
ispartof
issn
language eng
recordid cdi_arxiv_primary_1902_08341
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title FAVAE: Sequence Disentanglement using Information Bottleneck Principle
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-05T10%3A56%3A49IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=FAVAE:%20Sequence%20Disentanglement%20using%20Information%20Bottleneck%20Principle&rft.au=Yamada,%20Masanori&rft.date=2019-02-21&rft_id=info:doi/10.48550/arxiv.1902.08341&rft_dat=%3Carxiv_GOX%3E1902_08341%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true