A Scalable Second Order Method for Ill-Conditioned Matrix Completion from Few Samples
Christian Kümmerle, Claudio M. Verdun
Abstract
We propose an iterative algorithm for low-rank matrix completion with that can be interpreted as an iteratively reweighted least squares (IRLS) algorithm, a saddle-escaping smoothing Newton method or a variable metric proximal gradient method applied to a non-convex rank surrogate. It combines the favorable data-efficiency of previous IRLS approaches with an improved scalability by several orders of magnitude. We establish the first local convergence guarantee from a minimal number of samples for that class of algorithms, showing that the method attains a local quadratic convergence rate. Furthermore, we show that the linear systems to be solved are well-conditioned even for very ill-conditioned ground truth matrices. We provide extensive experiments, indicating that unlike many state-of-the-art approaches, our method is able to complete very ill-conditioned matrices with a condition number of up to $10^{10}$ from few samples, while being competitive in its scalability.
BibTeX
@InProceedings{pmlr-v139-kummerle21a,
title = {A Scalable Second Order Method for Ill-Conditioned Matrix Completion from Few Samples},
author = {K{\"u}mmerle, Christian and Verdun, Claudio M.},
booktitle = {Proceedings of the 38th International Conference on Machine Learning},
pages = {5872--5883},
year = {2021},
editor = {Meila, Marina and Zhang, Tong},
volume = {139},
series = {Proceedings of Machine Learning Research},
month = {18--24 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v139/kummerle21a/kummerle21a.pdf},
url = {https://proceedings.mlr.press/v139/kummerle21a.html},
abstract = {We propose an iterative algorithm for low-rank matrix completion with that can be interpreted as an iteratively reweighted least squares (IRLS) algorithm, a saddle-escaping smoothing Newton method or a variable metric proximal gradient method applied to a non-convex rank surrogate. It combines the favorable data-efficiency of previous IRLS approaches with an improved scalability by several orders of magnitude. We establish the first local convergence guarantee from a minimal number of samples for that class of algorithms, showing that the method attains a local quadratic convergence rate. Furthermore, we show that the linear systems to be solved are well-conditioned even for very ill-conditioned ground truth matrices. We provide extensive experiments, indicating that unlike many state-of-the-art approaches, our method is able to complete very ill-conditioned matrices with a condition number of up to $10^{10}$ from few samples, while being competitive in its scalability.}
}