Dictionary Learning for Blind One Bit Compressed Sensing

TL;DR

This paper introduces a dictionary learning algorithm tailored for blind one bit compressed sensing, enabling the reconstruction of sparse signals from one bit measurements by learning an unknown domain.

Contribution

It proposes a novel convex cost function and a steepest-descent method for learning the measurement and domain matrices in blind one bit compressed sensing.

Findings

The algorithm improves reconstruction accuracy over non-dictionary learning methods.

Performance increases with more training signals and sign measurements.

Experimental results validate the effectiveness of the proposed approach.

Abstract

This letter proposes a dictionary learning algorithm for blind one bit compressed sensing. In the blind one bit compressed sensing framework, the original signal to be reconstructed from one bit linear random measurements is sparse in an unknown domain. In this context, the multiplication of measurement matrix $\Ab$ and sparse domain matrix $\Phi$, \ie $\Db=\Ab\Phi$, should be learned. Hence, we use dictionary learning to train this matrix. Towards that end, an appropriate continuous convex cost function is suggested for one bit compressed sensing and a simple steepest-descent method is exploited to learn the rows of the matrix $\Db$. Experimental results show the effectiveness of the proposed algorithm against the case of no dictionary learning, specially with increasing the number of training signals and the number of sign measurements.