A bilinear algorithm for sparse representations (Q2475611)

A bilinear algorithm for sparse representation is proposed. The problem is to find a matrix \(A\) such that the known matrix \(X\) can be represented as \(X = AS\), where \(X \in \mathbb R^{m\times N}\), \ \(A \in \mathbb R^{m\times n}\) and \(S\in {\mathbb R}^{n\times N}\), assuming that all \(m \times m\) submatrices of \(A\) are nonsingular and that \(S\) is sparse (with at most \(m-1\) nonzero elements). The authors show that the problem can be reduced to a hyperplane clustering problem. Then they propose a bilinear algorithm for solving this problem which is robust to outliers, as it is checked in a simulation example.