| HAL-Inria | Publications, software ... of Inria's scientists |
Conference papers
Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms
Abstract : We propose a novel class of algorithms for low rank matrix completion. Our approach builds on novel penalty functions on the singular values of the low rank matrix. By exploiting a mixture model representation of this penalty, we show that a suitably chosen set of latent variables enables to derive an Expectation-Maximization algorithm to obtain a Maximum A Posteriori estimate of the completed low rank matrix. The resulting algorithm is an iterative soft-thresholded algorithm which iteratively adapts the shrinkage coefficients associated to the singular values. The algorithm is simple to implement and can scale to large matrices. We provide numerical comparisons between our approach and recent alternatives showing the interest of the proposed approach for low rank matrix completion.
Cited literature [2 references]
https://hal.inria.fr/hal-01025508
Contributor : Adrien Todeschini <>
Submitted on : Tuesday, July 22, 2014 - 8:14:02 AM
Last modification on : Thursday, February 11, 2021 - 2:36:03 PM
Long-term archiving on: : Monday, November 24, 2014 - 7:15:55 PM
Contributor : Adrien Todeschini <>
Submitted on : Tuesday, July 22, 2014 - 8:14:02 AM
Last modification on : Thursday, February 11, 2021 - 2:36:03 PM
Long-term archiving on: : Monday, November 24, 2014 - 7:15:55 PM