Data-sparse matrix computations Lecture 25: Low Rank + Sparse Matrix Recovery
نویسندگان
چکیده
In the previous lecture, we observed that it is possible to recover a sparse solution to Ax = b by solving a minimization problem involving the 1-norm. In this lecture, we consider a matrix A that can be written as A = L+ S, where L is a low rank matrix and S is a sparse matrix, and seek a method that recovers L and S. We remark that Lecture 26 forms a sequal to these notes and addresses the technical details related to recovery under the assumption that only a subset of the entries of A are observable. To motivate this work, we begin with two application-based examples.
منابع مشابه
Data Sparse Matrix Computations - Lecture 4
domain ΩT domain ΩS "well-separated" xi yi Specifically, let’s say we have domains ΩS of N source points and ΩT of M target points, and these domains are “well-separated” (we will formalize this in section 3). Our goal is to compute the influence of all source points onto target points. Let the M × N matrix [K]ij = K(xi, yj) and assume it is approximately low-rank, so that K ≈ UV T with U of si...
متن کاملBackground Recovery by Fixed-rank Robust Principal Component Analysis
Background recovery is a very important theme in computer vision applications. Recent research shows that robust principal component analysis (RPCA) is a promising approach for solving problems such as noise removal, video background modeling, and removal of shadows and specularity. RPCA utilizes the fact that the background is common in multiple views of a scene, and attempts to decompose the ...
متن کاملRank-Sparsity Incoherence for Matrix Decomposition
Suppose we are given a matrix that is formed by adding an unknown sparse matrix to an unknown low-rank matrix. Our goal is to decompose the given matrix into its sparse and low-rank components. Such a problem arises in a number of applications in model and system identification, and is NP-hard in general. In this paper we consider a convex optimization formulation to splitting the specified mat...
متن کاملFast Automatic Background Extraction via Robust PCA
Recent years have seen an explosion of interest in applications of sparse signal recovery and low rank matrix completion, due in part to the compelling use of the nuclear norm as a convex proxy for matrix rank. In some cases, minimizing the nuclear norm is equivalent to minimizing the rank of a matrix, and can lead to exact recovery of the underlying rank structure, see [Faz02, RFP10] for backg...
متن کاملSubspace based low rank & joint sparse matrix recovery
We consider the recovery of a low rank and jointly sparse matrix from under sampled measurements of its columns. This problem is highly relevant in the recovery of dynamic MRI data with high spatio-temporal resolution, where each column of the matrix corresponds to a frame in the image time series; the matrix is highly low-rank since the frames are highly correlated. Similarly the non-zero loca...
متن کامل