Show simple item record

hal.structure.identifierNVIDIA [NVIDIA]
dc.contributor.authorBojarski, Mariusz*
hal.structure.identifierDepartment of Electrical and Computer Engineering
hal.structure.identifierPolytechnic institute of New York University [NYU-Poly]
dc.contributor.authorChoromanska, Anna*
hal.structure.identifier
dc.contributor.authorChoromanski, Krzysztof*
hal.structure.identifierIndustrial Engineering and Operations Research Department [IEOR Dept]
dc.contributor.authorFagan, Francois*
hal.structure.identifierLaboratoire d'Intégration des Systèmes et des Technologies [LIST]
dc.contributor.authorGouy-Pailler, Cédric
HAL ID: 6827
ORCID: 0000-0003-1298-7845
*
hal.structure.identifierLaboratoire d'analyse et modélisation de systèmes pour l'aide à la décision [LAMSADE]
dc.contributor.authorMorvan, Anne*
hal.structure.identifierIndustrial Engineering and Operations Research Department [IEOR Dept]
dc.contributor.authorSakr, Nourhan*
hal.structure.identifier
dc.contributor.authorSarlos, Tamas*
hal.structure.identifierLaboratoire d'analyse et modélisation de systèmes pour l'aide à la décision [LAMSADE]
dc.contributor.authorAtif, Jamal
HAL ID: 15689
*
dc.date.accessioned2017-04-11T12:58:55Z
dc.date.available2017-04-11T12:58:55Z
dc.date.issued2017
dc.identifier.urihttps://basepub.dauphine.fr/handle/123456789/16505
dc.descriptionJMLR: W&CP, vol. 54en
dc.language.isoenen
dc.subjectMachine learningen
dc.subjectrandom projectionsen
dc.subject.ddc006.3en
dc.titleStructured adaptive and random spinners for fast machine learning computationsen
dc.typeCommunication / Conférence
dc.description.abstractenWe consider an efficient computational framework for speeding up several machine learning algorithms with almost no loss of accuracy. The proposed framework relies on projections via structured matrices that we call Structured Spinners, which are formed as products of three structured matrix-blocks that incorporate rotations. The approach is highly generic, i.e. i) structured matrices under consideration can either be fully-randomized or learned, ii) our structured family contains as special cases all previously considered structured schemes, iii) the setting extends to the non-linear case where the projections are followed by non-linear functions, and iv) the method finds numerous applications including kernel approximations via random feature maps, dimensionality reduction algorithms, new fast cross-polytope LSH techniques, deep learning, convex optimization algorithms via Newton sketches, quantization with random projection trees, and more. The proposed framework comes with theoretical guarantees characterizing the capacity of the structured model in reference to its unstructured counterpart and is based on a general theoretical principle that we describe in the paper. As a consequence of our theoretical analysis, we provide the first theoretical guarantees for one of the most efficient existing LSH algorithms based on the HD3HD2HD1 structured matrix [Andoni et al., 2015]. The exhaustive experimental evaluation confirms the accuracy and efficiency of structured spinners for a variety of different applications.en
dc.identifier.citationpages1020-1029en
dc.relation.ispartoftitleProceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS 2017)en
dc.relation.ispartofeditorSingh, Aarti
dc.relation.ispartofeditorZhu, Jerry
dc.relation.ispartofpublnameJMLR: Workshop and Conference Proceedingsen
dc.relation.ispartofdate2017
dc.relation.ispartofpages1568en
dc.identifier.urlsitehttp://proceedings.mlr.press/v54/en
dc.contributor.countryeditoruniversityotherUNITED STATES
dc.contributor.countryeditoruniversityotherFRANCE
dc.subject.ddclabelIntelligence artificielleen
dc.relation.conftitle20th International Conference on Artificial Intelligence and Statistics (AISTATS 2017)en
dc.relation.confdate2017-04
dc.relation.confcityFort Lauderdale, FLen
dc.relation.confcountryUnited Statesen
dc.relation.forthcomingnonen
dc.description.ssrncandidatenonen
dc.description.halcandidateouien
dc.description.readershiprechercheen
dc.description.audienceInternationalen
dc.relation.Isversionofjnlpeerreviewednonen
dc.relation.Isversionofjnlpeerreviewednonen
dc.date.updated2017-04-11T12:22:34Z
hal.identifierhal-01505587*
hal.version1*
hal.update.actionupdateMetadata*
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut
hal.author.functionaut


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record