Compressed Sensing with Sparse Binary Matrices: Instance Optimal Error Guarantees in Near-Optimal Time
For researchers in compressed sensing and streaming algorithms, this work provides a near-optimal runtime recovery algorithm with strong error guarantees, though it is an incremental improvement over existing sublinear-time methods.
The paper develops a compressed sensing method with sparse binary matrices that achieves instance optimal error guarantees and runs in O((k log k) log N) time, matching a lower bound up to a log factor. The method uses a new class of matrices constructed by subsampling rows from incoherent matrices, reducing the number of random bits needed.
A compressed sensing method consists of a rectangular measurement matrix, $M \in \mathbbm{R}^{m \times N}$ with $m \ll N$, together with an associated recovery algorithm, $\mathcal{A}: \mathbbm{R}^m \rightarrow \mathbbm{R}^N$. Compressed sensing methods aim to construct a high quality approximation to any given input vector ${\bf x} \in \mathbbm{R}^N$ using only $M {\bf x} \in \mathbbm{R}^m$ as input. In particular, we focus herein on instance optimal nonlinear approximation error bounds for $M$ and $\mathcal{A}$ of the form $ \| {\bf x} - \mathcal{A} (M {\bf x}) \|_p \leq \| {\bf x} - {\bf x}^{\rm opt}_k \|_p + C k^{1/p - 1/q} \| {\bf x} - {\bf x}^{\rm opt}_k \|_q$ for ${\bf x} \in \mathbbm{R}^N$, where ${\bf x}^{\rm opt}_k$ is the best possible $k$-term approximation to ${\bf x}$. In this paper we develop a compressed sensing method whose associated recovery algorithm, $\mathcal{A}$, runs in $O((k \log k) \log N)$-time, matching a lower bound up to a $O(\log k)$ factor. This runtime is obtained by using a new class of sparse binary compressed sensing matrices of near optimal size in combination with sublinear-time recovery techniques motivated by sketching algorithms for high-volume data streams. The new class of matrices is constructed by randomly subsampling rows from well-chosen incoherent matrix constructions which already have a sub-linear number of rows. As a consequence, fewer random bits than previously required are needed in order to select the rows utilized by the fast reconstruction algorithms considered herein.