Parallel computing heuristics for low-rank matrix completion

Charlie Hubbard, Chinmay Hegde

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    Current algorithms for low-rank matrix completion often suffer from scalability issues - both in terms of memory as well as running time - when presented with very large datasets. In this paper, we introduce new parallel computing heuristics that can greatly accelerate matrix completion algorithms when used in GPU-based computing environments. Our heuristics enable speeding up popular algorithms for nonlinear matrix completion on standard real-world test datasets by orders of magnitude, while being highly memory-efficient.

    Original languageEnglish (US)
    Title of host publication2017 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017 - Proceedings
    PublisherInstitute of Electrical and Electronics Engineers Inc.
    Pages764-768
    Number of pages5
    ISBN (Electronic)9781509059904
    DOIs
    StatePublished - Mar 7 2018
    Event5th IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017 - Montreal, Canada
    Duration: Nov 14 2017Nov 16 2017

    Publication series

    Name2017 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017 - Proceedings
    Volume2018-January

    Other

    Other5th IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017
    Country/TerritoryCanada
    CityMontreal
    Period11/14/1711/16/17

    ASJC Scopus subject areas

    • Information Systems
    • Signal Processing

    Fingerprint

    Dive into the research topics of 'Parallel computing heuristics for low-rank matrix completion'. Together they form a unique fingerprint.

    Cite this