Inverse finite-size scaling for high-dimensional significance analysis

dc.contributorAalto-yliopistofi
dc.contributorAalto Universityen
dc.contributor.authorXu, Yingying
dc.contributor.authorPuranen, Santeri
dc.contributor.authorCorander, Jukka
dc.contributor.authorKabashima, Yoshiyuki
dc.contributor.departmentDepartment of Computer Science
dc.contributor.departmentUniversity of Helsinki
dc.contributor.departmentTokyo Institute of Technology
dc.date.accessioned2018-08-01T13:35:00Z
dc.date.available2018-08-01T13:35:00Z
dc.date.issued2018-06-06
dc.description.abstractWe propose an efficient procedure for significance determination in high-dimensional dependence learning based on surrogate data testing, termed inverse finite-size scaling (IFSS). The IFSS method is based on our discovery of a universal scaling property of random matrices which enables inference about signal behavior from much smaller scale surrogate data than the dimensionality of the original data. As a motivating example, we demonstrate the procedure for ultra-high-dimensional Potts models with order of 1010 parameters. IFSS reduces the computational effort of the data-testing procedure by several orders of magnitude, making it very efficient for practical purposes. This approach thus holds considerable potential for generalization to other types of complex models.en
dc.description.versionPeer revieweden
dc.format.extent1-9
dc.format.mimetypeapplication/pdf
dc.identifier.citationXu , Y , Puranen , S , Corander , J & Kabashima , Y 2018 , ' Inverse finite-size scaling for high-dimensional significance analysis ' , Physical Review E , vol. 97 , no. 6 , 062112 , pp. 1-9 . https://doi.org/10.1103/PhysRevE.97.062112en
dc.identifier.doi10.1103/PhysRevE.97.062112
dc.identifier.issn2470-0045
dc.identifier.issn1550-2376
dc.identifier.otherPURE UUID: da24f220-7efd-4d55-9feb-3aabae209764
dc.identifier.otherPURE ITEMURL: https://research.aalto.fi/en/publications/da24f220-7efd-4d55-9feb-3aabae209764
dc.identifier.otherPURE LINK: http://www.scopus.com/inward/record.url?scp=85048212207&partnerID=8YFLogxK
dc.identifier.otherPURE FILEURL: https://research.aalto.fi/files/26472204/PhysRevE.97.062112.pdf
dc.identifier.urihttps://aaltodoc.aalto.fi/handle/123456789/32926
dc.identifier.urnURN:NBN:fi:aalto-201808014327
dc.language.isoenen
dc.relation.ispartofseriesPhysical Review Een
dc.relation.ispartofseriesVolume 97, issue 6en
dc.rightsopenAccessen
dc.titleInverse finite-size scaling for high-dimensional significance analysisen
dc.typeA1 Alkuperäisartikkeli tieteellisessä aikakauslehdessäfi
dc.type.versionpublishedVersion
Files