Compressive sensing accurately reconstructs a signal that is sparse in some basis from measurements, generally consisting of the signal’s inner products with Gaussian random vec...
The paper introduces a new framework for feature learning in classification motivated by information theory. We first systematically study the information structure and present a n...