Models in which the covariance matrix has the structure of a sparse matrix plus a low rank perturbation are ubiquitous in machine learning applications. It is often desirable for learning algorithms to take advantage of such structures, avoiding costly matrix computations that often require cubic time and quadratic storage. This is often accomplished by performing operations that maintain such structures, e.g. matrix inversion via the Sherman-Morrison-Woodbury formula. In this paper we consider the matrix square root and inverse square root operations. Given a low rank perturbation to a matrix, we argue that a low-rank approximate correction to the (inverse) square root exists. We do so by establishing a geometric decay bound on the true correction's eigenvalues. We then proceed to frame the correction has the solution of an algebraic Ricatti equation, and discuss how a low-rank solution to that equation can be computed. We analyze the approximation error incurred when approximately solving the algebraic Ricatti equation, providing spectral and Frobenius norm forward and backward error bounds. Finally, we describe several applications of our algorithms, and demonstrate their utility in numerical experiments.
翻译:共变矩阵具有稀薄矩阵结构的模型, 加上低级别扰动的模型, 在机器学习应用程序中无处不在。 学习算法往往可取于学习算法, 以利用这种结构, 避免往往需要立方时和二次存储的昂贵矩阵计算。 这通常通过执行维持这种结构的操作来实现, 例如通过谢尔曼- 莫里森- Woodbury 公式进行矩阵倒置。 在本文中, 我们考虑矩阵平方根和反平方根操作。 鉴于矩阵的级别低, 我们争论说, 存在对( 反向) 平方根的低排序近似修正。 我们这样做的办法是在真正校正的平方根值上建立几何度衰变曲线。 然后我们着手制定校正的校正框架, 并讨论如何计算出该方程式的低位解决方案。 我们分析了在大约解决平方位里加提方程式时发生的近似错误, 提供光谱和 Frobenius 规范前向和后向误差框。 最后, 我们描述了我们数运算中的几种应用。