Approximation of Positive Semidefinite Matrices Using the Nyström Method
Positive semidefinite matrices arise in a variety of fields, including statistics, signal processing, and machine learning. Unfortunately, when these matrices are high-dimensional and/or must be operated upon many times, expensive calculations such as the spectral decomposition quickly become a comp...
Saved in:
| Main Author: | |
|---|---|
| Format: | Dissertation |
| Language: | English |
| Published: |
ProQuest Dissertations & Theses
01.01.2011
|
| Subjects: | |
| ISBN: | 1124736786, 9781124736785 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Abstract | Positive semidefinite matrices arise in a variety of fields, including statistics, signal processing, and machine learning. Unfortunately, when these matrices are high-dimensional and/or must be operated upon many times, expensive calculations such as the spectral decomposition quickly become a computational bottleneck. A common alternative is to replace the original positive semidefinite matrices with low-rank approximations whose spectral decompositions can be more easily computed. In this thesis, we develop approaches based on the Nyström method, which approximates a positive semidefinite matrix using a data-dependent orthogonal projection. As the Nyström approximation is conditioned on a given principal submatrix of its argument, it essentially recasts low-rank approximation as a subset selection problem. We begin by deriving the Nyström approximation and developing a number of fundamental results, including new characterizations of its spectral properties and approximation error. We then address the problem of subset selection through a study of randomized sampling algorithms. We provide new bounds for the approximation error under uniformly random sampling, as well as bounds for two new data-dependent sampling methods. We continue by extending these results to random positive definite matrices, deriving statistics for the approximation error of matrices with Wishart and beta distributions, as well as for a broader class of orthogonally invariant and residual independent matrices. Once this theoretical foundation has been established, we turn to practical applications of Nyström methods. We explore new exact and approximate sampling methods for randomized subset selection, and develop greedy approaches for subset optimization. We conclude by developing the Nyström approximation as a low-rank covariance estimator that provides for computationally efficient spectral analysis while shrinking the eigenvalues of the sample covariance. After deriving expressions for its bias and mean squared error, we illustrate the effectiveness of the Nyström covariance estimator through empirical examples in adaptive beamforming and image denoising. |
|---|---|
| AbstractList | Positive semidefinite matrices arise in a variety of fields, including statistics, signal processing, and machine learning. Unfortunately, when these matrices are high-dimensional and/or must be operated upon many times, expensive calculations such as the spectral decomposition quickly become a computational bottleneck. A common alternative is to replace the original positive semidefinite matrices with low-rank approximations whose spectral decompositions can be more easily computed. In this thesis, we develop approaches based on the Nyström method, which approximates a positive semidefinite matrix using a data-dependent orthogonal projection. As the Nyström approximation is conditioned on a given principal submatrix of its argument, it essentially recasts low-rank approximation as a subset selection problem. We begin by deriving the Nyström approximation and developing a number of fundamental results, including new characterizations of its spectral properties and approximation error. We then address the problem of subset selection through a study of randomized sampling algorithms. We provide new bounds for the approximation error under uniformly random sampling, as well as bounds for two new data-dependent sampling methods. We continue by extending these results to random positive definite matrices, deriving statistics for the approximation error of matrices with Wishart and beta distributions, as well as for a broader class of orthogonally invariant and residual independent matrices. Once this theoretical foundation has been established, we turn to practical applications of Nyström methods. We explore new exact and approximate sampling methods for randomized subset selection, and develop greedy approaches for subset optimization. We conclude by developing the Nyström approximation as a low-rank covariance estimator that provides for computationally efficient spectral analysis while shrinking the eigenvalues of the sample covariance. After deriving expressions for its bias and mean squared error, we illustrate the effectiveness of the Nyström covariance estimator through empirical examples in adaptive beamforming and image denoising. |
| Author | Arcolano, Nicholas Francis |
| Author_xml | – sequence: 1 givenname: Nicholas surname: Arcolano middlename: Francis fullname: Arcolano, Nicholas Francis |
| BookMark | eNotj8tKAzEYRgMqaGvfIbgfyG1yWZbipdCqYF2XJPPHRp2kTlLRF_MF-mIO6OrsvvOdCTpNOcEJmlDKhOJSaXmOZqVERwgxnBPBLtByvt8P-Sv2tsaccA74MZdY4yfgJ-hjByGmWAGvbR2ih4KfS0wvuO4A33-XOhx_eryGusvdJToL9r3A7J9TtLm53izumtXD7XIxXzVvQtJGE8872mqplSKtUU5aRy0DcMZ6zzRzmhnayhGqc04okN4GrUInxoZg-BRd_c2Otz8OUOr2NR-GNBq3Whmieasp_wXAD0rn |
| ContentType | Dissertation |
| Copyright | Database copyright ProQuest LLC; ProQuest does not claim copyright in the individual underlying works. |
| Copyright_xml | – notice: Database copyright ProQuest LLC; ProQuest does not claim copyright in the individual underlying works. |
| DBID | 050 053 0BH 0H~ AAFGM ABUWG ADZZV AFKRA AGAJT AQTIP AZQEC BENPR CBPLH CCPQU DWQXO EU9 G20 M8- PHGZM PHGZT PIMPY PKEHL PQCXX PQEST PQQKQ PQUKI PRINS |
| DatabaseName | Dissertations & Theses Europe Full Text: Health & Medicine Dissertations & Theses Europe Full Text: Science & Technology ProQuest Dissertations and Theses Professional Dissertations & Theses @ Harvard University ProQuest Central Korea - hybrid linking ProQuest Central (Alumni) ProQuest Central (Alumni) - hybrid linking ProQuest Central UK/Ireland ProQuest Central Essentials - hybrid linking ProQuest Women's & Gender Studies - hybrid linking ProQuest Central Essentials ProQuest Central ProQuest Dissertations & Theses Global: The Sciences and Engineering Collection ProQuest One ProQuest Central ProQuest Dissertations & Theses A&I ProQuest Dissertations & Theses Global ProQuest Dissertations and Theses A&I: The Sciences and Engineering Collection ProQuest Central Premium ProQuest One Academic (New) Publicly Available Content Database ProQuest One Academic Middle East (New) ProQuest Central - hybrid linking ProQuest One Academic Eastern Edition (DO NOT USE) ProQuest One Academic (retired) ProQuest One Academic UKI Edition ProQuest Central China |
| DatabaseTitle | Publicly Available Content Database Dissertations & Theses Europe Full Text: Health & Medicine ProQuest One Academic Middle East (New) ProQuest Central Essentials ProQuest One Academic Eastern Edition ProQuest Central (Alumni Edition) ProQuest One Community College ProQuest Dissertations & Theses: Open ProQuest Dissertations & Theses Global: The Sciences and Engineering Collection ProQuest Dissertations and Theses Professional ProQuest Central China ProQuest Dissertations and Theses A&I: The Sciences and Engineering Collection ProQuest Dissertations & Theses Global Dissertations & Theses Europe Full Text: Science & Technology ProQuest Central Dissertations & Theses @ Harvard University ProQuest One Academic UKI Edition ProQuest Central Korea ProQuest Central (New) ProQuest One Academic ProQuest Dissertations & Theses A&I ProQuest One Academic (New) |
| DatabaseTitleList | Publicly Available Content Database |
| Database_xml | – sequence: 1 dbid: PIMPY name: Publicly Available Content Database url: http://search.proquest.com/publiccontent sourceTypes: Aggregation Database |
| DeliveryMethod | fulltext_linktorsrc |
| ExternalDocumentID | 2407746981 |
| Genre | Dissertation/Thesis |
| GroupedDBID | 050 053 0BH 0H~ 8R4 8R5 ABUWG AFKRA AZQEC BENPR CBPLH CCPQU DWQXO EU9 G20 M8- PHGZM PHGZT PIMPY PKEHL PQEST PQQKQ PQUKI PRINS Q2X |
| ID | FETCH-LOGICAL-k461-80c3d15868770597b6ab1a2eeb9acc282b829156b827dbb47e6caf87fd4473f93 |
| IEDL.DBID | BENPR |
| ISBN | 1124736786 9781124736785 |
| IngestDate | Mon Jun 30 06:54:56 EDT 2025 |
| IsOpenAccess | true |
| IsPeerReviewed | false |
| IsScholarly | false |
| Language | English |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-k461-80c3d15868770597b6ab1a2eeb9acc282b829156b827dbb47e6caf87fd4473f93 |
| Notes | SourceType-Dissertations & Theses-1 ObjectType-Dissertation/Thesis-1 content type line 12 |
| OpenAccessLink | https://www.proquest.com/docview/879083581?pq-origsite=%requestingapplication% |
| PQID | 879083581 |
| PQPubID | 18750 |
| ParticipantIDs | proquest_journals_879083581 |
| PublicationCentury | 2000 |
| PublicationDate | 20110101 |
| PublicationDateYYYYMMDD | 2011-01-01 |
| PublicationDate_xml | – month: 01 year: 2011 text: 20110101 day: 01 |
| PublicationDecade | 2010 |
| PublicationYear | 2011 |
| Publisher | ProQuest Dissertations & Theses |
| Publisher_xml | – name: ProQuest Dissertations & Theses |
| SSID | ssib000933042 |
| Score | 1.5545822 |
| Snippet | Positive semidefinite matrices arise in a variety of fields, including statistics, signal processing, and machine learning. Unfortunately, when these matrices... |
| SourceID | proquest |
| SourceType | Aggregation Database |
| SubjectTerms | Applied Mathematics |
| Title | Approximation of Positive Semidefinite Matrices Using the Nyström Method |
| URI | https://www.proquest.com/docview/879083581 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1LSwMxEB5s60E8qKio1ZKD18VmN7tJTuKj1R66LNpDPZU8YZG22q1F_5h_wD9mst2KoHgRAiHkEiZh5ptH5gM4jUIdGZ6owGLDAxJLEkhLWBARI9ptFWuKRUk2QdOUDYc8q2pziqqscqUTS0Wtp8rHyM8Y5R4tMHz-9Bx40iifXK0YNGrQ8I3KSB0al500u_vFXXewgtDIaeak6vO0Wsc_lHBpWbpb_zzTNmxef8uo78CamexC78J3Cn_Nl98S0dSirKzNWhh0b8a5Njb3UBP1y_78pkBl4QByWBClb8V89vE-Rv2SWnoPBt3O4Oo2qDgTgkeSYGdvVKRxzBJGqQNOVCZCYhEaI7lQyrlXkoXcuWxuolpKQk2ihGXUauLEYXm0D_XJdGIOAMVWKxoyhmnkUEaChTDKwTvLJREOY8aH0FwJZVS9-2L0JZGjP3ebsLGMzfpxDPX57MWcwLpazPNi1qpusQW1m7DtVlmvnz18AhCBqzE |
| linkProvider | ProQuest |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMw1V1NS8MwGA5jExQPKirq_MhBj0XTpk16EFG3sbGtDN1ht5JPKLJN1zndf_LsH_CPmXStCIq3HYRACemh5E2fPO-bN-8DwKnnSk-FgXA0UqGDfY4drjF1PKzYxYXwJUEsE5sgUUQHg7BXAm_FXRibVllgYgbUcixsjPycktCyBYquHp8cKxplD1cLBY3Fqmir-Yvx2NLLVs2Y98x1G_X-bdPJRQWcBxwgA8jCk8inASXEMAvCA8YRc5XiIRPC-B-cuqHxacyDSM4xUYFgmhItMSaetqWXDOBXsGd6ZVC5qUe9u1-iA4bFmBfMRhDkZaWKvv8D87ONrLHxv6ZgE6zXvuULbIGSGm2D1rWtg_6aLC5dwrGGvSzzbKbgvRomUunEEmnYzdQHVAqztAhomC6M5ul08vE-hN1MOHsH9Jfx5bugPBqP1B6AvpaCuJQi4hkOFSDGlDDkVYccM8Og_X1QLWwQ5391Gn8Z4ODP0ROw2ux3O3GnFbWrYG0RhbbtEJSnk2d1BFbEbJqkk-N8AUEQL9len3hABGI |
| linkToPdf | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMw1V3PS8MwFA5jExEPKirq_JGDHsvWNm3Sg4g4h2WuFNxhnkqSJjBkm65zun9MvPuP-dK1IijedhAKpbSH0i_58r3Xl_chdOo6qasCX1raVoFFPEEsoQmzXKJ4sym9lNo8N5ugUcT6_SCuoPdyL4wpqyw5MSfqdCxNjrzBaGDUArMbuqiKiFvti8cnyxhImR-tpZvGYoR01PwForfsPGwB1GeO077uXd1YhcGA9UB8G8hZuqntMZ9RCiqDCp8LmztKiYBLCbGIYE4A8Q2caCoEocqXXDOqU0Koq00bJiD_GghyAjOsFofd-P6XTAEoGngcFgW_aDFVXns_-D9f1Nob__dzbKL11rc6gi1UUaNtFF6a_uivg8VmTDzWOM4r0mYK36nhIFV6YAQ27uauBCrDebkEBgWMo3k2nXy8DXE3N9TeQb1lvPkuqo7GI7WHsKdTSR3GbOoCqr7NuZIganUgCAdl7e2jeolHUsz2LPkC4-DPuydoFUBKbsOoU0dri-S0OQ5RdTp5VkdoRc6mg2xyXIwljJIlw_UJnRMNVA |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adissertation&rft.genre=dissertation&rft.title=Approximation+of+Positive+Semidefinite+Matrices+Using+the+Nystr%C3%B6m+Method&rft.DBID=050%3B053%3B0BH%3B0H%7E%3BAAFGM%3BABUWG%3BADZZV%3BAFKRA%3BAGAJT%3BAQTIP%3BAZQEC%3BBENPR%3BCBPLH%3BCCPQU%3BDWQXO%3BEU9%3BG20%3BM8-%3BPHGZM%3BPHGZT%3BPIMPY%3BPKEHL%3BPQCXX%3BPQEST%3BPQQKQ%3BPQUKI%3BPRINS&rft.PQPubID=18750&rft.au=Arcolano%2C+Nicholas+Francis&rft.date=2011-01-01&rft.pub=ProQuest+Dissertations+%26+Theses&rft.isbn=1124736786&rft.externalDBID=HAS_PDF_LINK&rft.externalDocID=2407746981 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=9781124736785/lc.gif&client=summon&freeimage=true |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=9781124736785/mc.gif&client=summon&freeimage=true |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=9781124736785/sc.gif&client=summon&freeimage=true |