Near-Optimal Compressive Binary Search

Matthew L. Malloy, Robert D. Nowak

We propose a simple modification to the recently proposed compressive binary search. The modification removes an unnecessary and suboptimal factor of log log n from the SNR requirement, making the procedure optimal (up to a small constant). Simulations show that the new procedure performs significantly better in practice as well. We also contrast this problem with the more well known problem of noisy binary search.

picture_as_pdf flag

Knowledge Graph

arrow_drop_up

Comments

Sign up or login to leave a comment