Provable Randomized Coordinate Descent for Matrix Completion

dc.contributor.authorCallahan, Matthew
dc.contributor.authorVu, Trung
dc.contributor.authorRaich, Raviv
dc.date.accessioned2024-04-10T19:05:50Z
dc.date.available2024-04-10T19:05:50Z
dc.date.issued2024-03-18
dc.descriptionICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Seoul, Republic of Korea, 14-19 April 2024.
dc.description.abstractLow-rank matrix completion, the process of estimating a low-rank matrix from a small subset of its entries, has many applications including collaborative filtering and system identification. Many algorithms have been considered to address this problem. Coordinate descent has been previously proposed to tackle scalability both in terms of runtime and space complexity. Due to the use of regularization in the method, the method provides no convergence guarantees. Additionally, the choice of the regularization parameter can significantly affect the algorithm performance. Here, we study a regularization-free randomized coordinate descent method that uses an efficient periodic refactorization to guarantee a linear convergence rate. To support the proposed algorithm, we provide an analysis of the algorithm asymptotic convergence rate alongside a per-iteration computational complexity analysis. Using numerical experiments, we verify the correctness of our analysis and illustrate the overall computation advantage of the proposed approach.
dc.description.sponsorshipThis work was partially funded by the Consortium for Nuclear Forensics under Department of Energy National Nuclear Security Administration award number DE-NA0004142
dc.description.urihttps://ieeexplore.ieee.org/abstract/document/10446340
dc.format.extent5 pages
dc.genreconference papers and proceedings
dc.identifierdoi:10.13016/m2bdt8-wnl2
dc.identifier.citationCallahan, Matthew, Trung Vu, and Raviv Raich. “Provable Randomized Coordinate Descent for Matrix Completion.” ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), April 2024, 9421–25. https://doi.org/10.1109/ICASSP48485.2024.10446340.
dc.identifier.urihttps://doi.org/10.1109/ICASSP48485.2024.10446340
dc.identifier.urihttp://hdl.handle.net/11603/33003
dc.language.isoen_US
dc.publisherIEEE
dc.relation.isAvailableAtThe University of Maryland, Baltimore County (UMBC)
dc.relation.ispartofUMBC Faculty Collection
dc.relation.ispartofUMBC Computer Science and Electrical Engineering Department
dc.subjectAsymptotic convergence analysis
dc.subjectcoordinate descent
dc.subjectIterative algorithms
dc.subjectmatrix completion
dc.subjectMinimization
dc.subjectrandomized methods
dc.subjectScalability
dc.subjectSignal processing
dc.subjectSignal processing algorithms
dc.subjectSystem identification
dc.titleProvable Randomized Coordinate Descent for Matrix Completion
dc.typeText

Files