arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1302.6768math.NAcs.LGcs.NAstat.ML
收藏

缺失条目的矩阵逼近与补全

Missing Entries Matrix Approximation and Completion

School of Electrical Engineering Tel Aviv University(电气工程学院 特拉维夫大学);School of Computer Science Tel Aviv University(计算机科学学院 特拉维夫大学)

Gil Shabat, Yaniv Shmueli, Amir Averbuch

AI总结:

针对仅部分条目已知的矩阵,提出一系列算法实现矩阵补全与逼近,支持低秩、核范数、谱范数等多种约束,并证明凸情形下全局收敛,无需参数且适用于图像重建及偏微分方程数据恢复。

摘要:

我们描述了当矩阵仅部分条目已知时,用于矩阵补全和矩阵逼近的几种算法。逼近约束可以是任何对于完整矩阵已知其近似解的约束。对于低秩逼近,类似的算法最近在文献中以不同名称出现。在这项工作中,我们引入了矩阵逼近的新定理,并表明这些算法可以扩展到处理不同的约束,例如核范数、谱范数、正交约束等,这些约束与低秩逼近不同。由于这些算法可以从优化的角度看待,我们讨论了它们在凸情形下收敛到全局解的问题。我们还讨论了最优步长,并表明它在每次迭代中是固定的。此外,推导出的矩阵补全流是鲁棒的,不需要任何参数。该矩阵补全流适用于不同的谱最小化问题,并可应用于物理、数学和电气工程问题,例如图像数据重建以及来自偏微分方程(如用于电磁波的亥姆霍兹方程)的数据重建。

英文摘要:

We describe several algorithms for matrix completion and matrix approximation when only some of its entries are known. The approximation constraint can be any whose approximated solution is known for the full matrix. For low rank approximations, similar algorithms appears recently in the literature under different names. In this work, we introduce new theorems for matrix approximation and show that these algorithms can be extended to handle different constraints such as nuclear norm, spectral norm, orthogonality constraints and more that are different than low rank approximations. As the algorithms can be viewed from an optimization point of view, we discuss their convergence to global solution for the convex case. We also discuss the optimal step size and show that it is fixed in each iteration. In addition, the derived matrix completion flow is robust and does not require any parameters. This matrix completion flow is applicable to different spectral minimizations and can be applied to physics, mathematics and electrical engineering problems such as data reconstruction of images and data coming from PDEs such as Helmholtz equation used for electromagnetic waves.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑