Self-Concordant Analysis of Frank-Wolfe Algorithms
Projection-free optimization via different variants of the Frank-Wolfe (FW), a.k.a. Conditional Gradient method has become one of the cornerstones in optimization for machine learning since in many cases the linear minimization oracle is much cheaper to implement than projections and some sparsity n...
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| 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 | Dvurechensky, Pavel Ostroukhov, Petr Safin, Kamil Shtern, Shimrit Staudigl, Mathias |
| description | Projection-free optimization via different variants of the Frank-Wolfe (FW),
a.k.a. Conditional Gradient method has become one of the cornerstones in
optimization for machine learning since in many cases the linear minimization
oracle is much cheaper to implement than projections and some sparsity needs to
be preserved. In a number of applications, e.g. Poisson inverse problems or
quantum state tomography, the loss is given by a self-concordant (SC) function
having unbounded curvature, implying absence of theoretical guarantees for the
existing FW methods. We use the theory of SC functions to provide a new
adaptive step size for FW methods and prove global convergence rate O(1/k)
after k iterations. If the problem admits a stronger local linear minimization
oracle, we construct a novel FW method with linear convergence rate for SC
functions. |
| doi_str_mv | 10.48550/arxiv.2002.04320 |
| format | Article |
| fullrecord | <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2002_04320</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2002_04320</sourcerecordid><originalsourceid>FETCH-LOGICAL-a670-26ffa07139d8563233168e485516dcf92c3c50a8082be125068afbd713fe0cfc3</originalsourceid><addsrcrecordid>eNotzrFuwjAUhWEvDBX0ATqRF3B6fY2NGaOotEhIHUBijC6OL0SYuHJQVd6-BTqd6T_6hHhRUM6cMfBK-af7LhEAS5hphCeBmxBZ1qn3KbfUX4qqp3gduqFIXCwz9Se5S5FDUcVDyt3leB4mYsQUh_D8v2OxXb5t6w-5_nxf1dVakp2DRMtMMFd60TpjNWqtrAs3hrKt5wV67Q2QA4f7oNCAdcT79i_gAJ69Hovp4_aObr5yd6Z8bW745o7Xv1MLPb8</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Self-Concordant Analysis of Frank-Wolfe Algorithms</title><source>arXiv.org</source><creator>Dvurechensky, Pavel ; Ostroukhov, Petr ; Safin, Kamil ; Shtern, Shimrit ; Staudigl, Mathias</creator><creatorcontrib>Dvurechensky, Pavel ; Ostroukhov, Petr ; Safin, Kamil ; Shtern, Shimrit ; Staudigl, Mathias</creatorcontrib><description>Projection-free optimization via different variants of the Frank-Wolfe (FW),
a.k.a. Conditional Gradient method has become one of the cornerstones in
optimization for machine learning since in many cases the linear minimization
oracle is much cheaper to implement than projections and some sparsity needs to
be preserved. In a number of applications, e.g. Poisson inverse problems or
quantum state tomography, the loss is given by a self-concordant (SC) function
having unbounded curvature, implying absence of theoretical guarantees for the
existing FW methods. We use the theory of SC functions to provide a new
adaptive step size for FW methods and prove global convergence rate O(1/k)
after k iterations. If the problem admits a stronger local linear minimization
oracle, we construct a novel FW method with linear convergence rate for SC
functions.</description><identifier>DOI: 10.48550/arxiv.2002.04320</identifier><language>eng</language><subject>Computer Science - Learning ; Mathematics - Optimization and Control ; Statistics - Computation</subject><creationdate>2020-02</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/2002.04320$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2002.04320$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Dvurechensky, Pavel</creatorcontrib><creatorcontrib>Ostroukhov, Petr</creatorcontrib><creatorcontrib>Safin, Kamil</creatorcontrib><creatorcontrib>Shtern, Shimrit</creatorcontrib><creatorcontrib>Staudigl, Mathias</creatorcontrib><title>Self-Concordant Analysis of Frank-Wolfe Algorithms</title><description>Projection-free optimization via different variants of the Frank-Wolfe (FW),
a.k.a. Conditional Gradient method has become one of the cornerstones in
optimization for machine learning since in many cases the linear minimization
oracle is much cheaper to implement than projections and some sparsity needs to
be preserved. In a number of applications, e.g. Poisson inverse problems or
quantum state tomography, the loss is given by a self-concordant (SC) function
having unbounded curvature, implying absence of theoretical guarantees for the
existing FW methods. We use the theory of SC functions to provide a new
adaptive step size for FW methods and prove global convergence rate O(1/k)
after k iterations. If the problem admits a stronger local linear minimization
oracle, we construct a novel FW method with linear convergence rate for SC
functions.</description><subject>Computer Science - Learning</subject><subject>Mathematics - Optimization and Control</subject><subject>Statistics - Computation</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotzrFuwjAUhWEvDBX0ATqRF3B6fY2NGaOotEhIHUBijC6OL0SYuHJQVd6-BTqd6T_6hHhRUM6cMfBK-af7LhEAS5hphCeBmxBZ1qn3KbfUX4qqp3gduqFIXCwz9Se5S5FDUcVDyt3leB4mYsQUh_D8v2OxXb5t6w-5_nxf1dVakp2DRMtMMFd60TpjNWqtrAs3hrKt5wV67Q2QA4f7oNCAdcT79i_gAJ69Hovp4_aObr5yd6Z8bW745o7Xv1MLPb8</recordid><startdate>20200211</startdate><enddate>20200211</enddate><creator>Dvurechensky, Pavel</creator><creator>Ostroukhov, Petr</creator><creator>Safin, Kamil</creator><creator>Shtern, Shimrit</creator><creator>Staudigl, Mathias</creator><scope>AKY</scope><scope>AKZ</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20200211</creationdate><title>Self-Concordant Analysis of Frank-Wolfe Algorithms</title><author>Dvurechensky, Pavel ; Ostroukhov, Petr ; Safin, Kamil ; Shtern, Shimrit ; Staudigl, Mathias</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a670-26ffa07139d8563233168e485516dcf92c3c50a8082be125068afbd713fe0cfc3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Computer Science - Learning</topic><topic>Mathematics - Optimization and Control</topic><topic>Statistics - Computation</topic><toplevel>online_resources</toplevel><creatorcontrib>Dvurechensky, Pavel</creatorcontrib><creatorcontrib>Ostroukhov, Petr</creatorcontrib><creatorcontrib>Safin, Kamil</creatorcontrib><creatorcontrib>Shtern, Shimrit</creatorcontrib><creatorcontrib>Staudigl, Mathias</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Mathematics</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Dvurechensky, Pavel</au><au>Ostroukhov, Petr</au><au>Safin, Kamil</au><au>Shtern, Shimrit</au><au>Staudigl, Mathias</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Self-Concordant Analysis of Frank-Wolfe Algorithms</atitle><date>2020-02-11</date><risdate>2020</risdate><abstract>Projection-free optimization via different variants of the Frank-Wolfe (FW),
a.k.a. Conditional Gradient method has become one of the cornerstones in
optimization for machine learning since in many cases the linear minimization
oracle is much cheaper to implement than projections and some sparsity needs to
be preserved. In a number of applications, e.g. Poisson inverse problems or
quantum state tomography, the loss is given by a self-concordant (SC) function
having unbounded curvature, implying absence of theoretical guarantees for the
existing FW methods. We use the theory of SC functions to provide a new
adaptive step size for FW methods and prove global convergence rate O(1/k)
after k iterations. If the problem admits a stronger local linear minimization
oracle, we construct a novel FW method with linear convergence rate for SC
functions.</abstract><doi>10.48550/arxiv.2002.04320</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | DOI: 10.48550/arxiv.2002.04320 |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_arxiv_primary_2002_04320 |
| source | arXiv.org |
| subjects | Computer Science - Learning Mathematics - Optimization and Control Statistics - Computation |
| title | Self-Concordant Analysis of Frank-Wolfe Algorithms |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T14%3A27%3A41IST&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=Self-Concordant%20Analysis%20of%20Frank-Wolfe%20Algorithms&rft.au=Dvurechensky,%20Pavel&rft.date=2020-02-11&rft_id=info:doi/10.48550/arxiv.2002.04320&rft_dat=%3Carxiv_GOX%3E2002_04320%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 |